On SentinelAI, I engineered an industrial condition-monitoring platform with four ML analysis modules and 21 FastAPI endpoints. I trained a six-class fault classifier that achieved 94.0% accuracy and a 0.942 macro-F1 score on a blocked chronological test set.
I connected inference, decision logic, evidence provenance, retrieval, and maintenance reporting in an end-to-end workflow with PostgreSQL persistence. I also implemented a retrieval-bounded LangGraph maintenance copilot and validated the platform with backend, frontend, and Playwright tests.
For InsightPulse, I automated an ETL pipeline for financial news articles and FinBERT sentiment outputs into a SQLite analytics warehouse. I calculated company-level scores with explainable drivers and created Power BI dashboards for market trends, sentiment, and executive reporting.
I also built AlphaLens, integrating NewsAPI, FinBERT, and yFinance to generate explainable sentiment-based trading signals, displayed in an interactive Streamlit and Plotly dashboard. As Chapter Administrator for Microsoft Learn Student Ambassadors, I manage technical events and coordinate institute-wide internship programs.
