Role: Full Stack AI Engineer

Architected the AI core of an enterprise sustainability reporting platform. Built a Gemini AI-powered document intelligence pipeline that extracts structured data from 100+ compliance document types via custom prompt engineering and JSON schema validation. Engineered a cross-document synthesis engine (MultiDocQA) using LLM extraction with Redis distributed locking, a RAG chatbot with OpenAI embeddings, and an automated pipeline that maps extracted data directly into regulatory disclosure fields, eliminating weeks of manual data entry. Instrumented the entire LLM layer with Langfuse v4 observability, enabling end-to-end tracing of every OCR, classification, extraction, and multi-doc QA call in production.

Role: Flutter & AI Engineer

Award-winning AI platform built for rural Indian teachers managing multiple grade levels simultaneously in the same classroom. Engineered the Flutter mobile app and integrated a 4-agent AI system using Google's Agent Development Kit and Gemini API, with agents handling lesson planning, quiz generation, attendance via face recognition, and a multilingual help chatbot. Built a RAG pipeline backed by Firestore vector search over NCERT curriculum data, and used PaliGemma (vision-language model) for textbook photo scanning. Won the Breakthrough Concept Award at Google's Agentic AI Day (57,000+ participants) and Runner-Up at the Accel x Anthropic Dev Day Showcase.

Role: Flutter & Applied AI

Agentic AI platform serving India's 120M+ farmers with hyperlocal guidance through specialized AI models. Built the Flutter mobile app and contributed to the AI pipeline, featuring a QLoRA fine-tuned SmolVLM (2.2B params) for crop disease diagnosis from photos, TinyLlama models prefix-tuned on Kisan Call Centre transcripts for conversational farming advice, and an LSTM-based mandi price forecaster. Orchestrated by DHARTI, a master routing agent that classifies farmer intent and delegates to the right specialist. Also includes an offline SMS fallback via Twilio and an SOS distress detection system for farmer mental health.

Role: GenAI & Full Stack Engineer

Online Pathology Lab Booking & Reporting Platform built on a secure three-tier architecture. Architected an Agentic AI system powered by Gemini-2.5-flash with an Intent Classifier and Orchestrator that routes tasks to specialized sub-agents (Booking, Status, Report). Built AI-aided pathology interpretation delivering automated Gemini-generated summaries and recommendations to pathologists, reducing diagnostic backlog. Engineered automated PDF report/receipt generation, SMTP email notifications for key lifecycle events, and JWT-based role-based access control for patients, collectors, technicians, and admins. Deployed on Microsoft Azure VM (Linux) with systemd, Azure Blob Storage, and GitHub Actions CI/CD pipeline.