Raw transactional order data had no single view for leadership — revenue, product performance, and customer segments all lived in separate exports with no way to spot trends or compare periods.

Modeled the dataset in Power BI with a star schema, wrote DAX measures for YoY growth and KPI cards, and used Python for pre-cleaning and segmentation logic before loading it into the report.

Shipped an interactive executive dashboard covering the full order set with drill-down by product, region, and customer segment — turned into a reusable template for future reporting.

Vedic and Western astrology calculations are traditionally split across separate tools with dated interfaces, and neither offers AI-assisted interpretation of the results.

Built a PyQt6 desktop app combining both calculation systems in one place — Panchang, Kundali Milan — and wired in a local Ollama model so interpretations run fully offline.

A single desktop application with a holographic UI that merges traditional astrology logic with local AI, open-sourced on GitHub.

Recruiters manually re-read every resume against a job description, then hand off screener routing and interview scheduling as separate steps — slow and easy to bottleneck.

Built on the Lemma SDK to chain resume ingestion, automated JD fit scoring, screener routing, and interview scheduling into one pipeline instead of four disconnected tools.

An end-to-end AI co-pilot that takes a candidate from resume submission to a scheduled interview without manual hand-offs in between.