Project index / 2024 to 2026
A record of employer-owned systems, open-source work, research, and deployed experiments. Each entry states what can be inspected here or followed to a public artifact.
01 / INDUSTRY
LLM-powered WhatsApp inquiry platform for pharmaceutical distribution, built from scratch with Azure OpenAI classification, Azure App Service deployment, and Snowflake-backed logging.
Architecture
02 / INDUSTRY
FastAPI analytics stack delivering customer insights to pharmaceutical field representatives. A versioned SQLite snapshot strategy, validated in GitHub Actions, replaced recurring warehouse reads for the delivery use case.
Delivery architecture
03 / PRIVACY-AWARE ML
Local-first PII extraction and anonymization for sensitive banking text. The service pairs multilingual transformer inference with schema-based validation so detections can be cross-checked before redaction.
Protection architecture
04 / RESEARCH
A framework that separates forecasting from execution, combining LSTM models with PPO, DQN, and SAC agents. Evaluated using risk-adjusted NIFTY-50 backtests.
05 / DEPLOYED EXPERIMENT
A Random Forest classifier served through FastAPI to demonstrate the full path from a trained model to an inspectable prediction interface. The training data is synthetic by design.