AI Engineer building useful systems beyond the notebook. I work across LLM applications, agentic workflows, applied machine learning, and data products, with a focus on measurable outcomes, clear architecture, and maintainable delivery.
The stack I use to move from model idea to reliable product.
Languages & product engineering
AI, machine learning & orchestration
Retrieval, data & ingestion
APIs & delivery
MLOps, GPU & code quality
Selected work, with the engineering signal made explicit.
Open the project evidence log
| Build | Engineering signal |
|---|---|
| InsightBull | Hybrid FinBERT + Gemini financial sentiment platform with 91%+ accuracy, five data sources, and an automated market-hours pipeline Python · TypeScript · CSS · HTML |
| CortexaAI | Multi-agent prompt engineering across six LLM providers, a seven-node LangGraph workflow, SSE streaming, and 79 REST endpoints Python · TypeScript · FastAPI · React |
| VectraCite AI | Table-preserving ingestion over five real SEC filings (830 pages), hybrid BGE + BM25 retrieval with weighted RRF, adaptive cross-encoder reranking, and deterministic verification of every cited figure Python · FastAPI · Qdrant · Redis · Kubernetes |
| Retinal Vessel Segmentation | Nine-stage classical vision pipeline evaluated on 100 fundus images with 95.6% accuracy Python · OpenCV |
| Heart Disease Prediction | End-to-end modeling workflow with imbalance handling, model comparison, and a deployed Streamlit interface Python · Jupyter Notebook · Streamlit |
| Global Air Quality Dashboard | Interactive D3.js data story with six connected visualizations and a live GitHub Pages deployment JavaScript · HTML · CSS · D3.js |
Twelve months of building, measured rather than claimed.
I am most interested in the layer where model capability becomes product reliability: grounded retrieval, agent coordination, evaluation, observable pipelines, and interfaces people can actually use.
Built around evidence, not decoration.