Hi, I’m Nagharjun Mathi Mariappan 👋
Data Scientist focused on turning raw biomedical data into usable tools for clinicians. I design and deploy end to end machine learning systems from data pipelines and model training to containerised deployments on Docker on local and cloud based servers. Comfortable moving across the stack: Python and PyTorch for modelling, MLflow for experiment tracking, GitHub Actions for automation, and FastAPI for services.
Lately I have been exploring GPU programming, ML compilers, and model optimizations on hardware. I am experimenting with CUDA kernels, Triton, Apache TVM, MLIR/LLVM passes and containerised deployments.
I enjoy collaborating with physicians and engineers to bring models into everyday workflows under various settings. Constantly learning, open sourcing when I can and motivated by interesting projects.
Beyond my professional work, I enjoy playing soccer.
Latest Writing
Selected Projects
-
MCP‑Ollama Client – Offline chat client that merges tools from multiple Model‑Context‑Protocol servers into one local LLM.
-
Evading Ad‑Blockers – Poisoned object‑detection model that drops ad‑blocking recall 15 %.
-
NYC Station Sense – Parallel Spark pipeline that forecasts subway crowding with R² = 0.62.
Education
| Degree | School | Year | GPA |
|---|---|---|---|
| MS Computer Engineering | New York University | 2024 | 3.96/4.0 |
| B.Tech Electronics and Computer Engineering | Vellore Institute of Technology | 2022 | 8.67/10 |
Contact
- 📧 nagharjun2000@gmail.com
- đź‘” LinkedIn
- 🧑‍💻 GitHub
- 📞 +1 (917) 688‑8843