Intel® Student Ambassador Profiles
Current Ambassadors
Aaron Masuba, BRAC University
Aaron is developing the Applied Machine Learning and Industrial IoT (AMLIIoT) Architecture. This is a state-of-the-art system control suite based on intelligent edge electronic devices and sensor fusion for machine learning applications in industrial processes, automation, and diagnostics. He uses Intel® FPGA high-performance computing (HPC) applications to accelerate and scale operations. Aaron hopes to solve some of the world's most pressing challenges such as access to efficient, sustainable carbon-neutral energy and other control system applications where optimization is required through the oneAPI and Applied Machine Learning and Industrial IoT (AMLIIoT) architecture.
Adam Tuft, Durham University
Adam uses Intel® toolkits powered by oneAPI to investigate the scheduling behavior of task-based parallel programming runtimes used in HPC applications. He focuses on the efficient use of OpenMP* and DPC++ (the oneAPI implementation of SYCL*) for offloading to Intel® GPUs. Working within the ExCALIBUR task-parallelism research project, Adam uses the Intel toolkits to develop Otter, a performance analysis tool. Otter aims to help developers port existing code to use task-based parallelism effectively.
Addam Jensen, Ohio State University
Addam works with a team at Holocron Technologies that enables organizations to detect and analyze patterns in global events to drive confident decision-making at scale. Using machine learning, Holocron SITREP delivers real-time assessments of economic, political, and social indicators extracted from over 21.9 million news, social media, and digital sources collected per year. These breakthrough insights streamline information discovery and analysis into one workflow to deliver more rapid and thorough decisions at a moment’s notice.
Aditya Krishna, Sri Krishna College of Technology
Aditya is implementing a drug classification system using the random forest algorithm and optimizing it with oneAPI. This classification system for drugs based on their attributes achieves an accuracy of approximately 96.6%. This approach has potential in drug discovery, predicting drug interactions, and personalized medicine.
Aftab Ahmed, Sukkur IBA University
Aftab focuses on using AI techniques to build solutions for computer vision and digital image processing. He is working on a project that uses Intel® oneAPI Deep Neural Network Library (oneDNN), machine learning, deep learning techniques, and the Modified National Institute of Standards and Technology (MNIST) dataset. He is also part of multiple tech student communities on campus, and is a tech blogger and community builder.
Ahamed Thaiyub, KPR Institute of Engineering and Technology
Ahamed takes advantage of oneAPI technology to predict employee attrition, aiding organizations in reducing turnover costs. By analyzing historical data and performing feature engineering, he identifies key attrition indicators. The project encompasses data prep, feature engineering, model selection, training, optimization with oneAPI tools, and evaluation. Ahamed's work showcases the oneAPI role in data-driven human resources (HR) solutions for recruitment and retention.
Akshay Ramakrishnan, Sastra University
Akshay is focused on building powerful and precise machine learning systems for bioengineering applications and optimized healthcare systems using AI Tools from Intel. His project will use the capabilities of oneDNN, Intel® oneAPI Data Analytics Library (oneDAL), and other Intel technologies to enhance the performance and accuracy of machine learning models in healthcare.
Alper Şahıstan, University of Utah
Alper is adapting ZFP, a compressed format library for representing multidimensional floating-point and integer arrays, to work with SYCL through the Intel® DPC++ Compatibility Tool and other Intel profiling tools. ZFP's capacity to employ serial and parallel (OpenMP and CUDA*) compression techniques makes it an indispensable asset for applications that require efficient data exchange with storage systems. In this context, Alper uses the Intel DPC++ Compatibility Tool to facilitate the transition from CUDA to SYCL, and enhance and refine the implementation as needed. His plans involve exploring the oneAPI rendering capabilities for scientific volume visualization for ZFP compressed arrays.
Andrei Cozma, University of Tennessee, Knoxville
Andrei develops new ways to visualize and understand complex patterns and correlations in data. In a self-supervised manner, he uses deep neural networks to discover potentially hidden patterns and correlations, and uses them to generate meaningful visualizations about the underlying datasets. His work uses the AI Kit and oneDNN to develop and train neural networks. He also uses the Intel® Rendering Kit with Intel® OSPRay to create an interactive visualization application for the trained parameters.
Aneerban Saha, Manipal University Jaipur
Aneerban uses sentimental analysis and machine learning to develop an AI voice bot that listens to feelings that a user shares and then gives advice. He uses AI Tools from Intel for this project. His goal is to minimize problems such as depression, anxiety, and mental health issues.
Animikh Aich, Boston University
Animikh takes advantage of the power of OpenVINO and oneAPI toolkits to fine-tune and advance machine learning algorithms. Notably, he has successfully applied these tools in a prior endeavor, achieving face blurring with results achieving 60 FPS inference on 10th generation Intel Core i5 processors using OpenVINO. Presently, Animikh is immersed in the realm of multimodal learning. His goal is to construct a streamlined mini-LLaVA through the potent technique of knowledge distillation, finely tuned with oneAPI for optimal performance. This innovative approach not only ensures seamless compatibility across various hardware platforms but also significantly accelerates the inference process.
Ankur Singh, San Jose State University
Ankur is implementing Stable Diffusion* from scratch to make it more accessible to the masses. His primary focus is to use Intel-optimized libraries and packages for training as well as inference. He is not limiting himself to Stable Diffusion; he plans to implement DreamBooth, Text Inversion, and more. He also plans to develop an easy-to-use web application that anyone can use to play with Stable Diffusion and generate some cool looking images.
Arun GK, Christ University
Using machine learning, Arun's project aims to detect edible mushrooms based on their physical characteristics such as cap shape, cap color, gill size, spore print color, and habitat. Models such as logistic regression, decision trees, random forest, support vector machine (SVM), and XGBoost were used to build prediction models, and their accuracy was compared. A heat map was plotted to analyze the correlations between different features, and the relationships between other characteristics and edibility were explored. Arun used oneAPI and oneDAL for model optimization.
Bhaskar Trivedi, Madan Mohan Malaviya University of Technology
Bhaskar uses machine learning techniques to build applications for users in finance and healthcare. He uses AI Tools for a project that connects stock market traders. The project provides them with a no-code interface to predict the stock prices in real time and, through HPC, make better decisions.
Brett Bernstein, Colorado School of Mines
Brett uses AI to process and interpret geophysical data, makes use of excess data that exists in the industry, and extracts information from data that was once viewed as useless. He uses Intel-optimized libraries and oneDNN to create deep learning frameworks that can be adapted to a number of geophysical problems. Brett hopes to help educate members of the geophysics industry on topics like AI and deep learning using cutting-edge Intel technology.
Brishti Saha, Netaji Subhash Engineering College
Brishti is building faster and more secure movement for a bionic limb that can be controlled in various ways that depend on the user's preferences and skills. The electromyography (EMG) and inertial measurement unit (IMU) signals are received by a computer embedded in the prosthesis, which controls motors in the arm to create complex multidirectional movements simultaneously. To make faster and more secure movements, Brishti is creating a program that uses the OpenVINO™ toolkit and oneAPI for building the prosthesis models.
Chelsea Iluno, Prairie View A&M University
Chelsea implements, evaluates, and compares a regular decision tree model, a gradient-boosting decision tree, and Intel® Optimization for XGBoost* algorithm to predict house prices. In addition, she uses many other Intel-optimized frameworks, such as the Intel-optimized versions of Python* and TensorFlow* to enhance data preprocessing and analytics on the house price prediction. She aims to create a downloadable mobile device app to predict house and rent prices.
Daudi Wampamba, University of Nottingham
Daudi's project centers on the development of a parallel algorithm using oneAPI to find approximate solutions to the Quadratic Unconstrained Binary Optimization (QUBO), an NP-Hard minimization problem. His project compares this novel solution with existing quantum and classical approaches to the same problem, fostering a deeper understanding of their respective merits and limitations.
Deepak Joshi, Bharati Vidyapeeth's College of Engineering, Delhi
Deepak improves and optimizes existing machine learning models and develops crucial preprocessing pipelines to make them work efficiently and accurately on low-specification computers. He uses Intel toolkits (powered by oneAPI) to implement new machine learning models and extract meaningful data from huge datasets.
To hasten the process of training and predicting large data, Deepak uses oneAPI to take advantage of GPUs and CPUs as accelerators for his AI and natural language processing (NLP) projects.
Dev Aryan Khanna, Guru Gobind Singh Indraprastha University
Dev Aryan Khanna develops assistive smart glasses for blind people by taking advantage of the power of Intel technologies. The project uses AI Tools, oneDAL, and other Intel technologies to optimize object recognition, text identification, and navigation assistance that empower blind people with improved independence and accessibility.
Devesh Seethi, Northern Illinois University
Devesh builds multimodal frameworks on Intel® Developer Cloud using data from inertial sensors, vision, and audio to solve real-world problems to improve public health. He is creating a framework to track activity levels and behavioral patterns in Alzheimer patients using person re-identification, point of interest detection, motion tracking, and monocular depth estimation in the Intel® Distribution of OpenVINO™ toolkit.
N. Dharanidharan, Jansons Institute of Technology
N. Dharanidharan's project focuses on revolutionizing the prediction of genetic disorders using Intel toolkits, powered by oneAPI. By harnessing the immense computational power and advanced tools provided by Intel toolkits and oneDNN, he aims to deliver highly accurate predictions for a diverse range of genetic disorders. The successful implementation of the genetic disorder predictive model holds immense potential in transforming the healthcare industry and advancing genetic research.
Diego Abad, Florida A&M University and Florida State University (FAMU-FSU) College of Engineering
Diego's research project involves creating a basic neural network using C++, and using CUDA to modify the original C++ code into a CUDA format. He is using the Intel® DPC++ Compatibility Tool to translate the CUDA code into DPC++, and then modify the translated file to ensure it compiles while using the time library to measure the time it takes to complete the task.
Drake Du, Harvard University
Drake uses oneAPI technologies to build, optimize, and scale his programming projects. He uses the state-of-the-art machine learning algorithms contained in oneDAL to help develop a web application that helps university students to more effortlessly assess and appraise the inclusivity of student organizations on campus. Outside of serving as an Intel® Student Ambassador, Drake is involved with Harvard Tech for Social Good, the Harvard-MIT Mathematics Tournament (HMMT), and the Harvard Computer Society.
Ebenezer Daniel, DePaul University
Ebenezer's project focuses on using Allen's Mice Brain Data to study the sagittal and coronal data of genes in the brain stem that regulate orofacial behaviors. The dataset includes 20,000 genes, and their corresponding correlation tables contain 400 million entries. The intent is to identify patterns and reveal hidden relationships among them for multiple structures in the brain stem. This will help us take a crucial step toward uncovering the complex neural circuitry associated with orofacial behaviors. To achieve this task, we plan to use oneAPI to accelerate the correlation table calculations on CPUs and GPUs.
Fatih Şengül, Sivas University of Science and Technology
Fatih uses AI techniques to detect cyberattacks for intrusion detection systems (IDS). To do this, he uses oneDNN, Intel® Optimization for TensorFlow, AI Tools, and Intel Developer Cloud. Fatih uses computer vision and digital image processing techniques using the Intel Distribution of OpenVINO toolkit and OpenCV to detect surface cracks on roads and architectural constructions.
Fernando Schettini, SENAI CIMATEC
Fernando's project creates informative and interactive Jupyter Notebook scripts to provide an accessible platform for learning the fundamentals of perceptron in a neural network using Intel® Distribution for Python*. Additionally, the project covers essential concepts related to neural networks, search algorithms, HPC, and simulations.
Gagan Agarwal, Kalinga Institute of Industrial Technology
Gagan uses oneAPI to optimize quantum circuit simulation for composite quantum systems. He plans to use the Intel® HPC Toolkit to unlock quantum computing for heterogeneous devices that would otherwise be restricted to high-performance machines. It will help users to better understand the quantum behavior, optimizing the simulation with AI Tools.
Gangesh Basker, SASTRA University
Gangesh applies machine learning methods to develop models in computer vision to improve driver and passenger safety in autonomous vehicles. His project uses the capabilities of oneDNN and other Intel technologies to enhance the performance and accuracy of machine learning models in driver and passenger safety. This system reduces the number of accidents on the road by detecting the driver's drowsiness, warning them using an alarm, and assisting the driver using the front-end application.
Guan Yan Lye, University of Nottingham (Malaysia)
The Generic Circuit Level Tool for evaluation of Nano-Crossbar Memory using Memristors is a software tool designed to help researchers and engineers in the field of memristor technology. It is a circuit-level tool that automates the process of generating the memristor array circuit in a spice environment, making it easier and faster for researchers to design and simulate memristor circuits. The tool also allows researchers to easily modify and optimize the simulation circuit parameters, such as the size of the memristor array circuit or the type of memristor used.
D. Gopalakrishnan, Excel Engineering College
D. Gopalakrishnan is developing an NLP-based emotion recognition system that is used to understand accurate emotion of users. He used oneDNN, which provides optimized implementations of deep learning building blocks for deep learning applications and accelerate the speed of the training.
Hadrien Gayap, University of Moncton
Hadrien's project focuses on lung cancer diagnosis using deep learning. His project uses deep learning to improve the accuracy and sensitivity of early detection of lung cancer. According to a systematic review of conference papers and scientific journals published in 2022, deep learning algorithms have become a superior way to automatically diagnose disease.
Hamed Barzamini, Northern Illinois University
Hamed's project focuses on implementing requirement engineering for AI-enabled Software (RE-AIS), a framework that uses humans' semantic knowledge of domain concepts to establish a benchmark for evaluating the quality of randomly collected training datasets in AIS. For this project, Hamed uses the AI Kit to take advantage of the capabilities of oneDNN, oneDAL, and other Intel technologies. By incorporating these tools, he enhances the performance and accuracy of machine learning models in autonomous driving software systems.
Harish Raaghav DV, KGiSL
Harish's project is focused on classifying drugs based on patient characteristics such as age, sex, blood pressure, cholesterol level, and sodium-to-potassium ratio (Na_to_K). He employs various machine learning algorithms and techniques to achieve accurate drug classification.
Harshit Saini, Gautam Buddha University
Harshit developed an automatic labelling tool that streamlines the process of image annotation. This tool employs OpenCV for the user interface of the annotator and uses PyTorch*, which benefits from several back end libraries from the AI Kit. Its primary objective is to automate and expedite the labor-intensive task of image labelling. This back end from Intel enhances the tool's performance, making it highly adaptable to various hardware configurations. Harshit's vision includes further integration with the Intel Distribution for Python, creating a stand-alone Linux* version of the tool to bolster its capabilities for image annotation efficiency.
Harvey Johnson, University of Nottingham
Harvey is using oneAPI to implement an efficient and timely data preprocessing pipeline for AI. This system uses GPU and accelerator resources to compress and decompress hundreds of millions of data samples in seconds. This process simplifies loading large datasets for Gavin AI to allow for training on larger datasets without I/O or storage bottlenecks.
Hunter Price, University of Tennessee, Knoxville
Hunter uses the Intel Rendering Toolkit to build a scientific visualization service for geospatial data. He uses Intel OSPRay to render and serve terrain data using a C++ back end. This process allows all recourse-intensive computation to take place on the server side, which allows anyone with an internet connection to view scientific data. Hunter's goal is to make scientific visualization more accessible to those that do not have the resources to process large amounts of data.
Iancecil Njoroge, Strathmore University
Iancecil aims to develop a simple VR game using oneAPI, a software development environment for optimizing applications on multiple hardware platforms. The game will be optimized to run on both CPUs and GPUs, using the Intel Rendering Toolkit to enable specific rendering of images. His project will provide valuable resources and documentation for other developers interested in using oneAPI for VR game development.
Imed-Eddine Haouli, Badji Mokhtar University
Imed-Eddine uses deep learning techniques and medical imaging to develop new methods for automatically detecting different diseases, such as glaucoma and COVID-19. The methods are based on the new deep architecture, which is Vision Transformer (ViT) and use the Grad-Cam visualization technique to demonstrate the areas that influence the model's decision in order to increase the interpretability of the models. To do so, he uses oneDNN and OpenVINO™ integration with TensorFlow* to implement these methods.
James Hammer, University of Tennessee, Knoxville
James automates the deployment of Intel tools to Docker* in swarm mode and various scalable platforms with a focus on graphics and data visualization systems. He and his team finds novel uses for oneAPI technologies, such as developing a web-compliant front-end for Intel OSPRay Studio. James makes automated software deployment to HPC environments and enables the deployment of other oneAPI toolkits and development.
Jiaqi Wang, University of Washington
Jiaqi is developing a platform for AI project deployment that helps to integrate and optimize model usage and examination. He uses Intel toolkits to investigate AI modeling and to address cross-architecture compatibility. Also, he will be using Intel Distribution of OpenVINO toolkit and Intel Developer Cloud to help manage deployment.
Joel John Joseph, Christ University
Joel's project uses machine learning algorithms to provide customized crop recommendations to farmers based on their soil data. The project used oneDAL to optimize the performance of the machine learning models used in the application. This platform enabled faster computation times, improved accuracy, and enhanced efficiency, resulting in more accurate and reliable crop recommendations. The application used various machine learning models such as support vector machine (SVM), logistic regression, random forest (RF), and XGBoost to generate crop recommendations for farmers.
Joshua Shiells, University of Kent, United Kingdom
Joshua is using oneAPI to take advantage of GPUs as accelerators for his AI projects and research. He is working on "Gavin," which is a transformer chatbot trained on the Reddit* Pushshift dataset. He is also focused on building a machine learning library using oneAPI to easily support multiple applications on both GPU and CPU compute. He used oneAPI to simplify the amount of code needed to handle many different devices or accelerators.
Kaan Olgu, University of Bristol
Kaan is using oneAPI, specifically SYCL, to delve into the realm of random memory accesses on FPGAs, with a specific focus on optimizing the Breadth-First Search (BFS) algorithm. The goal of his research is to uncover strategies for enhancing the efficiency of random memory accesses in BFS implementations. By using the power of oneAPI's SYCL, he seeks to unlock new insights and techniques that can significantly improve memory handling in FPGA-based BFS computations. This endeavor holds promise for advancing the performance capabilities of BFS applications across a range of domains.
Kamesh Ragupathi, Jansons Institute of Technology
Kamesh's project is focused on building a time-efficient and approximate stock predictions for Google* stocks with a recurrent neural network (RNN) using the long short-term memory (LSTM) Keras model. The Intel Distribution of OpenVINO toolkit enables these models to be trained more efficiently, and results in faster training times. He uses the ADAM optimization algorithm and optimizes a mean squared error (MSE) loss function.
Keerthana K M, Manipal Academy of Higher Education
Keerthan's project detects anomalies in time series over a million rows dataset. She needs oneAPI to perform operations over this large dataset for better computation and high performance.
Keerthi Parthipan, Anna University Regional Campus Madurai
Keerthi's project addresses and provides a comprehensive machine learning and deep learning solution to common agricultural problems. By using the potential of convolutional neural networks (CNN), the project achieves disease identification in tomato leaves using image analysis and offers insights by visualizing trends in local tomato markets. The project uses Intel Optimization for TensorFlow to accelerate computation and enhance accuracy.
Kenichi Hayakawa, Costa Rica Institute of Technology
Kenichi's project uses oneAPI toolkits and Intel FPGA development tools to analyze where hardware and software performance bottlenecks lie when running computationally intensive workloads on multicore architectures. His goal is to gain insights into the hardware and software aspects that limit performance, and then design and implement efficient hardware-accelerated solutions for tasks such as cryptography, image processing, and algorithm acceleration, while ensuring that the performance improvements achieved do not compromise portability.
Lakshay Taneja, Shoolini University
Lakshay develops new and improves existing machine learning models and processes to make models work in a production environment. Lakshay focuses on developing machine learning algorithms to extract meaningful information from large amounts of datasets. He uses Intel toolkits to implement new machine learning models and improve the existing datasets.
Manas Marwah, Bharati Vidyapeeth's College of Engineering
Manas develops new ways to visualize and understand complex patterns in data. His work includes the Intel Rendering Toolkit, including ray-surface hit testing, volumetric space iteration, and image denoise. He also uses the Intel Rendering Toolkit with Intel OSPRay to create an interactive visualization application.
Manikanta Bukapindi, Dayananda Sagar College Of Engineering
Manikanta uses AI Tools and machine learning to develop an AI system that detects real-world objects like pedestrians, cars, and obstacles, program autonomous driving, and then detect violations. His goal is to minimize problems caused due to violations and make it easier for autonomous driving systems to make faster and better decisions in real-world scenarios.
Martin Muchai, University of Nairobi
Martin's project focuses on developing a oneAPI-powered AI chatbot that can effectively handle customer queries, provide personalized recommendations, and enhance the overall customer experience using oneAPI libraries. The proposed solution will use oneAPI's extensive array of libraries and tools to develop an AI chatbot that can understand natural language inputs and provide relevant responses.
Mayurdhvajsinh Jadeja, Marwadi University
Mayurdhvaj is developing a sign-language-to-text-and-speech converter for individuals who are deaf and mute. This innovative project uses advanced technologies, including real-time gesture recognition through the MediaPipe library, oneAPI libraries like oneDNN and Intel Distribution of OpenVINO toolkit, and NLP and speech-synthesis techniques. The system accurately detects and interprets sign language gestures, and then converts them into text and synthesized speech. By improving communication between deaf and mute individuals, Mayurdhvaj's project aims to enhance inclusivity and accessibility, benefiting millions in the deaf and mute community.
Melbin Martin, Christ University
Melbin's smart garbage segregation project aims to use AI and machine learning to efficiently and effectively sort waste into different categories, such as plastic and glass, using oneDNN. Image classification for recycling refers to the use of machine learning algorithms to automatically classify images of waste materials into their respective categories. This process involves training a model using a large dataset of labeled images, and then using this model to predict the category of new, unlabeled images.
Michael Martinez, University of Houston
Michael uses AI Tools and oneDNN to develop and train computer vision models that automatically detect different rock types and fabrics in digital images and remote sensing data. Human-based classification using these data types is prone to variance due to differing interpretations. Automating these workflows allows for fast, accurate processing of large amounts of geoscience data with less variance. Accurate models could build helpful tools for geoscientists in the field and at workstations.
Migara Amarasinghe, Florida A&M University and Florida State University (FAMU-FSU) College of Engineering
Migara focuses on investigating parallel architectures for AI algorithms. To conduct hardware performance analyses, he uses a diverse range of heterogeneous computer systems that consist of high-end GPUs and CPUs with single and multicore processing capabilities. To optimize the scalability of AI algorithms, Migara uses CUDA and high-level programming models such as SYCL. He is using the Intel DPC++ Compatibility Tool to migrate code from CUDA to SYCL for the Intel® oneAPI DPC++ Compiler and the Intel HPC Toolkit to maximize the performance of Intel processors.
Miguel Graça, INESC-ID
Miguel develops machine learning methods for epistasis detection, a bioinformatics application that aims at identifying associations between Single Nucleotide Polymorphisms and complex diseases. To this end, he is using the Intel-optimized versions of TensorFlow, as well as OpenVINO, to deploy enhanced transformer models for epistasis on a wide variety of AI accelerators.
Mohammed Bangie, University of the Witwatersrand
Mohammed's project harness the power of oneAPI, DPC++, and oneDNN. His goal is to enable hydroponic systems to autonomously adapt and make informed decisions, thereby eliminating the need for extensive labeled training data. He uses oneAPI toolkits to exploit the full potential of CPUs, GPUs, and FPGAs. By harnessing these diverse capabilities, he empowered his system to learn, adapt, and optimize in real time. His project seeks to maximize crop yield and resource utilization, ultimately transforming the landscape of hydroponic agriculture.
Mohit Sharma, University of Windsor
Mohit uses Intel Optimization for TensorFlow in the Intel AI Analytics Toolkit to speed up techniques for improving machine learning classification model results. He developed a method for using a hierarchy of binary classifiers for multiclass classification. This method calculates the probability score for belonging to a class by multiplying the results of multiple binary classifiers. These binary classifiers are trained on groups of classes with a similar number of data points.
Muhammad Faeez Shabbir, The Islamia University of Bahawalpur
Muhammad's Sentify project uses oneAPI to perform sentiment analysis on natural language text. The project involves processing input text through various NLP techniques, such as tokenization, stemming, and part-of-speech tagging to extract meaningful information about the text. The goal of the project is to provide a reliable and accurate way to analyze the sentiment of large volumes of text data, which can be used in a wide range of applications such as social media monitoring, customer feedback analysis, and market research.
Muhammad Hanzaila Maqsood, Technical University Munich
Muhammad uses deep learning to train the agent in Python by using multiple libraries and Intel® oneAPI Math Kernel Library (oneMKL) for mathematical functions like PyTorch and TensorFlow. He also uses oneDNN to increase GPU performance while using the deep learning neural network in PyTorch.
Nadav Schneider, Ben-Gurion University
Nadav researches and develops AI technologies focusing on computer vision and large-scale optimization domains. While studying for a master's degree, he is supervised by Dr. Gal Oren and Dr. Yuval Pinter, who are researching how to automatically convert a serial code into a distributed parallelism MPI code using generative language models.
Nejla Harris, University of North Carolina (UNCC) at Charlotte
Nejla's project aims to create murals as 3D occupiable spaces, which involves transforming flat surfaces into three-dimensional environments that can be experienced by people as they move through the space. This approach to mural-making will include scientific methods of solar architecture study. oneAPI technology can be used in various ways to enhance and facilitate a creative project. The Integrated Design Research Lab at UNCC uses optimized code from Intel oneAPI to run code across multiple architectures, such as CPUs and GPUs.
Nicholas Synovic, Loyola University Chicago
Nicholas develops new and improves existing computer vision models to run on low powered systems. He uses the AI Kit to develop and test low-powered machine learning models. This involves developing power-efficient and performant pipelines to prepare, process, and render data on devices. Nicholas focuses on developing computer vision models and finding solutions that can enable other model classes to run efficiently.
Nishank Satish, Dayananda Sagar College of Engineering
Nishank focuses on applying machine learning methods to tackle real-world problems by using computer vision. He is working on a traffic management system that addresses traffic issues, which include emergency vehicle detection, dynamic traffic signaling, and violation detection. The dynamic traffic signaling and emergency vehicle detection models use the Intel® AI Analytics Kit. The OpenVINO Toolkit is being used to enhance object detection models.
Nitin Dantu, Northeastern University
Nitin uses AI Tools to build and fine-tune classical machine learning and deep learning algorithms. He works with machine learning on large-scale data for a variety of domains, including healthcare, finance, real estate, agricultural technology (agro tech), and autonomous vehicles. Using Intel GPUs, Nitin creates robust, and scalable machine learning pipelines that are suitable for production in real time.
Orlando Mota, SENAI CIMATEC
Orlando's project uses oneAPI, which is pivotal for studying the DPC++ programming language and its application in scientific computing. He uses reverse time migration (RTM) as a proof-of-concept to assess DPC++ performance and flexibility. oneAPI also helps evaluate the DPC++ ability to handle basic data structures and algorithms. Its tools assist in generating precise documentation at each development stage.
Owen McGrath, Illinois Institute of Technology
Owen explores the oneAPI programming framework, specifically focusing on DPC++ and oneTBB, and their suitability and performance for very fined-grained parallelism, where each task may consist of only a few instructions. He runs benchmarks on oneTBB and compares the results to other parallel programming solutions.
Padmakumar RP, PSNA College of Engineering & Technology
Padmakumar's project focuses on developing an efficient object-detection deep learning model for autonomous vehicles with a strong focus on using oneAPI libraries. Using AI Tools, he optimizes the model to achieve exceptional performance in real-time processing. He achieves low latency and enhanced training and inference speed by incorporating the Intel® Extension for PyTorch* and using Intel® Neural Compressor libraries.
Polly Ren, University of Chicago Illinois
Polly uses machine learning techniques to build a university-specific course search tool based on students' inputted interests and desired career paths. This project could be useful for students who may be hesitant to ask these questions to their advisors and are unable to find specific information online. Her work uses oneDAL and oneDNN from the oneAPI specifications.
Poornima Nookala, Illinois Institute of Technology
Poornima is using Intel® VTune™ Profiler, Intel® Trace Analyzer, and Intel® Inspector in her projects in analyzing the performance bottlenecks and optimizing the runtime to scale up to hundreds of cores on modern many-core architectures. Her goal is to reduce overheads of tasking in parallel runtime systems as a step towards exascale computing.
Prajwal Kumar, Maharshi Dayanand University
Prajwal uses oneDNN and Intel Optimization for TensorFlow to optimize various charging stations for electronic vehicles (EV). His project involves developing a new EV charging network to eliminate inconveniences due to a small charging infrastructure and long charging times—the primary obstacles to mobility decarbonization.
Praveen Kumar, KGISL
Praveen developed a resume screening model using oneDAL library to classify resumes into predefined categories. After importing libraries, he performed exploratory data analysis (EDA) and cleaned data with text preprocessing. Using the power of oneDAL, he built and evaluated nine models, including k-nearest neighbor, linear support vector, Stochastic Gradient Descent, and more. Using Term Frequency-Inverse Document Frequency (TF-IDF) for text transformation, he achieved high accuracy. Stochastic Gradient Descent emerged as the top-performing model, attaining 100% accuracy. The oneDAL library significantly enhanced performance and efficiency in this project.
Rachel Selina Rajarathnam, The University of Texas at Austin
Rachel works on methods and algorithms to accelerate the process of FPGA circuit placement for improving design scalability. Her DREAMPlaceFPGA, an accelerated global placement framework that's implemented using a deep-learning toolkit, uses CPUs and GPUs. The AI support in oneAPI provides a suitable platform for further development and extension of the DREAMPlaceFPGA by allowing the use of CPUs, GPUs and FPGAs in one platform.
Raghul Senthilkumar, Amrita Vishwa Vidyapeetham
Raghul is building a real-time face recognition system using oneDNN and the AI Kit. His current focus is on building an automated attendance system that can be used in different scenarios. Raghul tests different models using the AI Kit to provide fast and efficient feedback to the application. He also uses various oneAPI models to improve the accuracy and reliability of machine learning models in healthcare.
Raison Sabu, Christ University
Raison's project aims at using AI and machine learning to efficiently sort waste into different categories using oneDNN. This process involves training a model using a large dataset of labeled images and then using this model to predict the category of new, unlabeled images. The goal of image classification for recycling is to improve the efficiency and accuracy of recycling processes by automating the sorting of materials, reducing human error, and increasing the amount of recyclable materials that can be recovered.
Re'em Harel, Ben-Gurion University
Re'em develops and integrates scientific applications in HPC systems using MPI+X paradigms, focusing on MPI and OpenMP on multicore and heterogeneous architectures. He is also developing a scalable framework benchmark named ScaleSALE and implementing OpenMP schemas to numerical schemas. In addition, Re'em is researching AI techniques for OpenMP directives that can be incorporated in the Intel® Advisor.
Rick Mondal, Narula Institute of Technology
Rick uses oneAPI in his project to accelerate the training and inference of large language models (LLMs). LLMs are computationally expensive to train and deploy, so it is important to use efficient hardware and software tools. oneAPI allows him to target a variety of hardware architectures, including CPUs, GPUs, and FPGAs, with a single code base.
Rohan Sethi, Loyola University Chicago
Rohan is focused on optimizing computer vision models to run on low-power edge devices. He uses the AI support in the AI Kit to implement and test the efficiency of data preprocessing pipelines on such devices. Rohan plans on developing and comparing performance of generative and discriminative machine learning pipelines to assess optimal solutions for healthcare and nonhealthcare computer vision applications.
Sage Lyon, University of Massachusetts Lowell
Sage is performing interoperability testing and hardening of the latest hardware technologies and open source software. He is focused on cloud computing, 5G networking, and virtualized radio access network (vRAN) use cases. For performance benchmarking, Sage uses the oneAPI-based open RAN reference architecture, FlexRAN.
Sai Rama Raju Penmatsa, San Jose State University
Raju focuses on developing models in computer vision, and is working on improving road safety by enabling timely response from medical responders. His project involves using Intel Distribution of OpenVINO toolkit for detection. Also, he is currently doing a research project on using vehicle-to-everything (V2X) communications to enable coperception among multiple ego vehicles, which would help in making better decisions in autonomous vehicles.
Sapphira Ching, University of Pennsylvania
Sapphira has led several student organizations at the University of Pennsylvania. She is advocating oneAPI on campus and attracting other talented students to join the Intel Student Ambassador program for oneAPI.
Shivaram Velayutham, Indian Institute of Technology, Madras
Shivaram's project implements a bitonic sorting algorithm using oneAPI and explores its optimization possibilities for efficient running on modern heterogeneous computing architectures. The bitonic sorting algorithm is particularly well suited for parallel processing on shared-memory systems, making it an ideal candidate for acceleration on GPUs, FPGAs, and other accelerators.
Shubham Luharuka, RV Institute of Technology and Management*
Shubham is developing an optimized and efficient task-oriented deep neural network to enhance the quality of videos. He uses Intel Optimization for TensorFlow and oneAPI to target the FPGA circuit for its real-time implementation. To reduce the training time of the model, he uses the Intel Developer Cloud. This implementation can be used in areas such as medical simulation, image processing, material engineering, video processing, computational materials science, optical flow estimation, computational electromagnetics, and more.
Sinda Besrour, University of Moncton
Sinda's bird species classification is a deep learning project based on oneAPI and TensorFlow. The data includes a total of 1,856 audio records split into 40 bird species. She implemented the project on the Intel Developer Cloud and created a conda* environment that inherits from the already existing TensorFlow environment.
Skye Hart, Colorado School of Mines
Skye uses machine learning methods to integrate and explore large geoscientific datasets to discover hidden relationships that could help to discover new mineral exploration targets. She also uses machine learning methods for inverting geophysical data. For this work, she uses Intel toolkits.
Subash Palvel, Jansons Institute of Technology
Subash works on music generation using oneDNN, drawing attention to the exciting fusion of music and AI. He describes how the AI music generator (powered by oneDNN) transforms the traditional music composition process. By training LSTM neural networks on an extensive MIDI dataset, the project produces personalized soundtracks that can elevate any production scenario. The technology automates the composition process, ensures efficiency, and unleashes creativity.
Subhadip Saha, JIS College of Engineering
Subhadip's project focuses on using the power of machine learning and sentimental analysis to develop an AI bot that can understand and respond to a user's emotions. He plans to use AI Tools for this purpose. His objective is to alleviate prevalent issues such as depression, anxiety, and mental health concerns.
Taimur Muhammad Khan, National University of Computer and Engineering Sciences, Islamabad
Taimur develops a spell correction model for Roman Urdu using the Noisy Channel model to correct nonword errors. He uses Intel toolkits to implement optimized and efficient data preprocessing techniques. Taimur uses the Intel Developer Cloud to access corpus and reduce training time. This process streamlines the data flow and helps load large corpora with ease.
Tal Kadosh, Ben-Gurion University
Tal uses the latest advancements in the field of large language models related to code to develop a method for automatically generating and inserting OpenMP pragma into serial code. By exploiting different code representations, Tal's method improves the accuracy of OpenMP parallelization detection and can potentially accelerate the implementation of computationally intensive tasks. This method can be incorporated into the Intel Advisor tool.
Taufeq Razakh, University of Southern California
As a member of the Collaboratory for Advanced Computing and Simulations (CACS) group, Taufeq leads performance optimization of nonadiabatic quantum molecular dynamics (NAQMD) and neural-network quantum molecular dynamics (NNQMD) simulation engines for Intel architectures. His operations are on HPC processors and accelerators from Intel, which results in Taufeq keeping up with the most recent advancements in oneAPI to extract the best performance during his application development cycles.
Tushar Suman, Poornima Group of Institutions
Tushar's project aims at drawing astrological charts using algorithms and calculations from Vedic math in DPC++. Using the OpenVINO toolkit, interpreting astrological charts would be faster, more accurate, and more user friendly by displaying the charts in a 3D view.
Utsab Khakurel, Howard University
Using predictive models, Utsab’s project uses the English Premier League dataset from seasons 2021/2022 and 2022/2023 to assess and validate the optimal model for predicting match winners. The data undergoes a comprehensive preprocessing phase, encompassing tasks such as data cleaning, addressing missing values, substituting absent numerical values with medians, categorization, label encoding, normalization, and employing SMOTE for target label balancing to ensure equitable predictions. Utsab uses k-nearest neighbors (KNN), Naïve Bayes, and decision tree classifiers to train and make predictions. Intel® Extension for Scikit-learn* is used to enhance model accuracy and performance.
Utsav Mehta, Pandit Deendayal Energy University
Utsav is focused on automating and innovating firmware using AI. He is currently working on a project to create an automated trash ecosystem using machine learning and oneDNN.
Vedansh Jaiswal, Shri Ramdeobaba College of Engineering and Management
Vedansh is working on a machine learning project that uses facial expression recognition to determine a user's mood and then suggests various song playlists based on that mood. His project works by obtaining a live video feed from a webcam and running it through the model to forecast emotion. He uses oneDAL for a better user experience and to make it possible for blind people to listen to music that suits their mood.
Vignesh Nagavel, VSB Engineering College, Karur
Vignesh is developing a hybrid model with recurrent neural networks to estimate river flow through hydropower plants from pluviometer and satellite precipitation data. For this work, Vignesh uses oneDNN, Intel Optimization for TensorFlow, and the Intel Developer Cloud.
Wonjae Choi, Georgia Institute of Technology
Wonjae focuses on machine learning optimization and how to analyze data for single-cell RNA-sequence data and other types of data on a single-cell level. He is working on knowing cellular mechanisms during differentiation, disease progression, and the development of cells. Wonjae is also using AI to create a cloud-based platform for solving common problems in finance, healthcare, and reducing the digital divide.
Yehonatan Fridman, Ben-Gurion University
Yehonatan researches the recoverability of scientific applications in HPC systems using the non-volatile RAM (NVRAM) technology, focusing on the Intel® Optane™ Persistent Memory product. Yehonatan is specifically interested in implementing recoverable mechanisms in OpenMP to enable concurrent algorithms running with OpenMP to run reliably.
Youssef Faqir, Complutense University of Madrid
Youssef uses Intel toolkits to develop portable algorithms without losing performance for multiarchitecture devices such as CPUs and GPUs. For that work, he developed the Non-Negative Matrix Factorization (NMF) algorithm in DPC++ (the oneAPI implementation of SYCL) and OpenMP, and the K-means algorithm focused to run over multivendor accelerators. Youssef plans to extend these works to FPGAs and compare the differences among architectures.
Yuri Winche Achermann, RWTH Aachen University
Yuri works on a predictive maintenance system for the manufacturing industry. In particular, he checks metal-cutting tool wear during each cycle of machining. He does this by using computer vision based on a deep learning algorithm that segments the wear of the tool image to approve (or not) an extended use for a tool's lifetime optimization. At the same time, the system keeps the data and shows dashboards of tool wearing for engineering insights.
Project
Zhibo Li, University of Edinburgh
Zhibo is working on a high-performance declarative data collection system based on oneAPI. He developed a front end for the data collection using C++ metaprogramming. Based on the front end, he is integrating concurrent data structures from oneTBB as back ends and applying DPC++ (SYCL) to parallelize the program, and to port it to multiple platforms.
Become a oneAPI Student Ambassador
Get ready to inspire and influence your peers from across the globe. Demonstrate your technology leadership in AI, HPC, visualization, and rendering.
Join the Student Community for oneAPI
For more information, contact us.