Summary
Boise State PhD researcher seeking Summer 2027 internships in ML engineering, applied AI research, ML systems, data science, and software engineering. Builds reproducible Python/PyTorch training, evaluation, inference-benchmarking, and deployment-verification workflows. Research focuses on power-of-two quantization, multiplier-free neural networks, LLM inference, computer vision, and edge AI. Expected graduation: December 2027.
Education
PhD in Computing, Data Science and Machine Learning
Boise State University · Advisor: Dr. Omiya Hassan
B.S. in Computer Science and Engineering
American International University-Bangladesh
Research Experience
Graduate Research Assistant · Boise State University, LPiNS Lab
- Build Python and PyTorch pipelines for preprocessing, training, evaluation, experiment orchestration, metric aggregation, and reporting.
- Develop controlled experiments across classification, vision, and language workloads with ablations, multi-seed evaluation, and error analysis.
- Create reusable workflows for validation, structured logging, checkpoint analysis, and model comparison.
Teaching Assistant · Boise State University
- Support algorithms, data structures, digital systems, AI hardware systems, programming, debugging, and problem solving.
Industry Experience
Software Engineer I · Playense
Software and Application Development
- Developed, tested, and debugged application features using Python, C++, and Dart while translating requirements into software.
- Implemented and integrated software components, investigated technical issues, and adapted solutions to evolving requirements.
- Completed and delivered eight client projects through final handoff.
Peer-Reviewed Publications
- I. A. Udoy and O. Hassan, “Multiplier-Free LLM Linear Layers via Weights-Only Power-of-Two QAT,” IEEE ICAD, 2026. DOI: 10.1109/ICAD69378.2026.11609079.
- M. I. Nadim et al., including I. A. Udoy, “Attention-Driven Hybrid Deep Learning for Automated Alzheimer’s Disease Severity Assessment via MRI Neuroimaging,” Emerging Science Journal, 10(3), 2026.
- I. A. Udoy, R. Sharmin, M. M. Hossain, S. K. Islam, and O. Hassan, “Lightweight Binarized Neural Network for Real-Time Sleep Apnea Detection on Edge Hardware,” IEEE MeMeA, 2025. DOI: 10.1109/MeMeA65319.2025.11068064.
- I. A. Udoy and O. Hassan, “AI-Driven Technology in Heart Failure Detection and Diagnosis: A Review of the Advancement in Personalized Healthcare,” Symmetry, 17(3), 469, 2025. DOI: 10.3390/sym17030469.
Preprint & Current Research
Preprint: M. M. Arifin et al., including I. A. Udoy, “A Survey on the Application of Generative Adversarial Networks in Cybersecurity,” arXiv:2407.08839, 2024.
Submitted manuscript, under review: SPARQ: Single-Term Power-of-Two Quantization for Efficient Neural Network Inference. Submitted to IEEE Transactions on Artificial Intelligence; not yet accepted or published.
Selected ML Systems and AI Projects
Po2QAT: Reproducible power-of-two quantization-aware training examples for compact CNN, ViT, and language models, with inspectable exports.
SCALE-Sim Analysis for SPARQ: Scheduling and memory-traffic studies across five compact model topologies. Results cover constrained weighted-subgraph simulation, not measured silicon or full-model speedups.
Model Export and Benchmarking Toolkit: Model export, cross-runtime output-parity checks, inference benchmarks, failure records, and deployment reports.
Technical Skills
Programming / DataPython, SQL, C/C++, Bash, pandas, NumPy, JupyterMachine Learningscikit-learn, XGBoost, PyTorch, TensorFlow, regression, classification, predictive modeling, time-series forecastingAutomationHugging Face, LLM applications, retrieval workflows, agentic AI, automated experimentationEngineeringGit/GitHub, Linux, Docker, pytest, FastAPI, REST APIs, SQLite, Streamlit, automated testing