Aspiring Machine Learning Engineer · Computer Science @ Texas Tech University · Break Through Tech AI Fellow
I'm a Computer Science student at Texas Tech University working toward a career as a Machine Learning Engineer. I build ML systems end to end, from data pipelines and model training to fast, GPU-accelerated inference.
As an Honors Undergraduate Research Scholar in the DISCL Lab, I research agentic AI pipelines for genomic variant interpretation and write CUDA kernels to speed up GPU inference. My goal is to build reliable, efficient ML tools that make a real difference in health care and science.
- AI Studio, Fall 2026, challenge project with Accenture. With a student team, building an NLP pipeline that detects 41 types of legal clauses in commercial contracts and ranks contracts by risk for human review.
- Implemented the DistilRoBERTa multi-label fine-tuning pipeline and the TF-IDF baseline. The fine-tuned model reached 0.71 macro F1 on benchmark categories, beating the baseline's 0.59.
- Project: ai-contract-risk-triage
- Researching agentic LLM pipelines for genomic variant interpretation.
- Writing CUDA kernels for GPU inference.
- Ran weekly HPC benchmarks, container testing, and CUDA-Q / PennyLane workflows on HPE Cray EX, Bridges-2, OLCF Odo, and IQM Garnet.
| Project | What it does | Tech |
|---|---|---|
| ai-contract-risk-triage | Detects risky clauses in contracts and ranks contracts for legal review (Break Through Tech AI Studio × Accenture) | PyTorch, Hugging Face, scikit-learn |
| ooi-agent | AI assistant in JupyterLab that finds, fetches, and plots ocean data from the Ocean Observatories Initiative | Python, LLM agents, Jupyter |
| mnist-neural-network | Neural networks from scratch in pure Python, including an MNIST digit classifier | Python, NumPy, Jupyter |
| animoo | Text-to-3D animation with NVIDIA Nemotron agents (HackUTD 2026) | FastAPI, three.js, generative AI |
| winterclassic | HPC benchmarking and quantum workflows across four supercomputers | Slurm, MPI, CUDA-Q |
| nand2tetris | A complete 16-bit computer built from NAND gates up, plus an 8-bit breadboard CPU | HDL, Assembly |
| Area | Skills |
|---|---|
| Machine Learning | PyTorch, TensorFlow, Hugging Face Transformers, scikit-learn, LangChain, ONNX Runtime |
| Data & Computer Vision | NumPy, pandas, OpenCV, MediaPipe, Jupyter |
| Languages | Python, C, C++, CUDA, Rust, Fortran, Assembly, JavaScript |
| HPC & Tools | MPI, OpenMP, OpenACC, CUDA-Q, Slurm, Docker, Git, Linux, HPE Cray EX |
- Model evaluation and calibration for imbalanced, multi-label data
- Efficient transformer inference: quantization and custom CUDA kernels
- Contributing to open-source ML libraries
The best way to reach me is by email at naeem.islam@ttu.edu or on LinkedIn. I'm open to ML engineering internships and research collaborations.


