AI/ML engineer · creative coder
I build practical ML systems end-to-end — data → model → API → UI.
I like clean metrics, fast APIs, and friendly interfaces.
- Bengali NLI — XLM-R (LoRA) for premise–hypothesis; 84% F1
- Doc Q&A (RAG) — 1k+ docs with SentenceTransformers · Chroma · LangChain · Gemini on AWS Lambda
- Emotion Drift Monitor — DistilBERT + ZenML/MLflow + Streamlit on YouTube comments
- Sales Call Analyzer — predicts conversion; drafts tailored follow-ups with GPT-4o
- ResNet-50 Classifier — Flask + Docker API; 87.6% test accuracy
- Dex (C++) — CLI for project scaffolding, Git automation, and workflow ops
PyTorch · Hugging Face Transformers · TensorFlow · scikit-learn
MLflow / ZenML · FastAPI / Flask · Docker · AWS (S3 / Lambda / EC2)
PostgreSQL · MongoDB · Redis · Vector DBs (Chroma)
React · Next.js · Node.js / TypeScript
- Shipping small, measurable features
- Keeping latency low and observability high
- Making ML products people actually want to use
Open an issue, start a discussion, or ping me—always up for smart builds and fun collabs.
Pushing ML to be useful, not just impressive.