DietDoctor AI
An AI dietician living inside WhatsApp — deterministic Python tools do the math, so it never miscounts a calorie.
AGENTIC AI × DATA ENGINEERING — GURUGRAM, IN
LLM agents, RAG and document intelligence — designed, shipped and running end-to-end on my own Kubernetes cluster, with production data pipelines underneath. The agent below happens to be one of them — ask it anything.
An AI dietician living inside WhatsApp — deterministic Python tools do the math, so it never miscounts a calorie.
Per-patient medical RAG — reads scanned and handwritten records, cites every answer, never leaks between patients.
An end-to-end streaming backbone into Snowflake with CDC — the class of pipeline I run in production.
CHECKING…··EMAIL ME
SHIPPED SYSTEMS
A personal AI dietician living entirely inside WhatsApp. Deterministic Python tools do the math — the LLM never miscounts a calorie.
Answers only from each patient's own records — Gemini vision OCR reads handwritten notes, hybrid retrieval cites every source. Zero cross-patient leaks.
Reads marksheets and IDs with GPT-4o vision, cross-checks identities, chases missing fields over email — and fills the form itself.
Ask a question in English; a Llama 3.3 ReAct agent writes and runs the Spark SQL over a Medallion lakehouse — autonomously.
An end-to-end streaming backbone with change-data-capture — the same class of pipeline she runs in production.