From IoT Circuits to AI Systems
I build things that turn messy data into clear decisions. Whether it's automating enrollment pipelines, deploying multi-agent healthcare systems, or wiring up IoT sensors for urban safety, I care about the full loop: real data, real stakeholders, real impact.
Every role taught me something different. Here's the path so far.
Where I learned to build for stakeholders, not just for code
Where I learned what "production-grade" actually means
RFID-authenticated coffee machine built with Raspberry Pi Pico. Users scan their card to pick from 4 coffee types. The system reads the card via SPI-based RC522, authenticates, and controls relay-driven dispensing. All MicroPython.
IoT system tackling two urban safety problems in Indian cities. Ultrasonic sensors track waste bin fill levels in real time, while gas sensors (methane, H₂S), water level detectors, and tilt sensors monitor manhole status. Open and toxic manholes cause hundreds of sanitary worker deaths every year. Automated GSM/Wi-Fi alerts reach municipal authorities before workers approach hazardous sites.
24-hour hackathon project, built with a team of 4. Autonomous robot with IR-based fire detection, obstacle avoidance using ultrasonic ranging, and DC motor driven movement. Won runner-up.
What I reach for when building.
Things I built because I wanted to understand how they work.
Production-style platform automating prior authorization, care gap detection & patient risk triage. 5 LangGraph + CrewAI agents, FastAPI backend (8 endpoints), 8-tab Streamlit dashboard, RAG over 63 clinical guidelines with cross-encoder reranking, and a 93-node clinical knowledge graph.
Live news feeds + 10 years of market data to assess portfolio downside under real-world crisis scenarios. Regime-switching tail risk framework using VaR, CVaR, tail dependence & Clayton copula modeling. Separate crash models for calm vs. crisis markets.
End-to-end pipeline processing 10M+ FDA recall records across 9,100+ devices. Custom Recall Pattern Severity Score (RPSS) across 5 risk dimensions. ML-based recall prediction + time series forecasting. Top 2% critical devices drive 60%+ of high severity recalls.
Academic Lean Six Sigma project applying the full DMAIC framework to a real-world problem: Tesla Model S batteries overheating during supercharging in extreme heat. Ran a 5-factor DOE analysis (ambient temperature, charging algorithm, cooling system, battery life cycle, safety procedures), identified root causes using fishbone and Pareto analysis, and proposed actionable improvements including supercharging restrictions for older batteries, alternative coolant fluids, and revised safety testing. Targeted a 40% reduction in overheating incidents.
Automated NLP pipeline using VADER to evaluate 10,000+ open-ended student feedback comments. Custom tuned classification thresholds for academic context. Enabled leadership to quantify qualitative feedback and prioritize institutional improvements at scale.
Architected a relational database from scratch to structure multi-source enrollment data. Normalized schema with SQL joins creating analytical views with derived columns. Connected to Tableau for interactive dashboards, reducing manual reporting effort by 30%.
M.S. in Industrial / Engineering Management
B.E. in Electronics & Communication Engineering
Project Management Institute
Council of Supply Chain Management Professionals