PwC India
AI/ML Intern
Shipped high-throughput document processing and NLP workflows used in production.
Built a document pipeline processing ≈1,800 pages/hour with OpenCV + YOLOv8 (≈92% precision), reducing manual data entry by ~95%.
Designed an NLP workflow for the Arunachal Pradesh Government: ingests monthly reports from 60+ departments, fine-tunes Llama-2, and produces dashboards; cut analysis time from ~2 days to ~30 minutes.
Authored an OSS licensing compliance checklist adopted firm-wide, reducing legal review cycles by ~40%.
Stack: Python, OpenCV, YOLOv8n, Hugging Face Transformers.
Orion Racing, K. J. Somaiya CoE
Data Acquisition Engineer
Built real-time telemetry and analytics for a Formula Student team.
Integrated 25+ CAN-Bus sensors and streamed telemetry to a React dashboard with sub-200ms latency (battery, thermal, suspension).
Automated anomaly detection across 100k+ datapoints per lap, accelerating fault diagnosis ~5×.
Created calibration & documentation pipelines that reduced rookie onboarding time by ~40%.
Stack: Python, C/C++, React, CAN-Bus.