AI/ML Engineer · Bengaluru, IN
I build document intelligence pipelines, fault-tolerant systems, and applied ML tools - from OCR that reads scanned invoices to a RAFT consensus cluster running a shared drawing canvas.
Built an internal OCR document-understanding pipeline in a team of two - CNN layout detection, transformer text recognition, attention-based field extraction - reaching 88% extraction accuracy on scanned documents.
Designed the preprocessing layer end-to-end: denoising, deskewing, adaptive thresholding across ingestion, a key contributor to accuracy on low-quality scans.
Trained XGBoost and k-NN models on librosa audio features and CNN image features to detect Parkinson's disease from voice, spiral, and wave-drawing inputs - 85% classification accuracy across three modalities.
Built a Flask app for real-time sample upload and AI symptom analysis, unifying all three input modes behind one interface.
State Champion, Zonal Runner-up & National Finalist - RBI90 National Quiz Competition (prize pool exceeding ₹6L); top-3 finishes across multiple inter-college and inter-university quizzes.
Multiple awards at inter-college Model United Nations conferences, including an Honourable Mention at Jigyasa 4.0 MUN; delegate across several committees.
Marketing & Operations Head, QQC Quiz Club, PES University - organized college and inter-college quizzes and curated question sets across science, tech, and current affairs.
Marketing Head, MUNSoc, PES University - managed end-to-end logistics for UniCon, an inter-college MUN conference with hundreds of participants.
Get in touch
Click to copy