Modeling + evals
Clear training loops, evaluation harnesses, and error analysis.
ML/AI engineer
Building practical ML/AI solutions for ambitious ideas.
I’m Tanmay Dogra — a full-stack ML/AI engineer who builds models, training pipelines, and production systems that turn research into real products. I specialize in agentic AI workflows, retrieval systems, and deployment patterns that make complex ideas reliable at scale. My focus is shipping practical AI that teams can trust and evolve.
Current focus
Agentic RAG systems
Designing retrieval pipelines and evaluation loops that keep AI outputs grounded.
Proof stack
How I ship reliable ML products
RAG + agents
Retrievers, chunking, orchestration, and human-in-the-loop checks.
Production ML
Inference services, monitoring, reliability, and scaling patterns.
Product delivery
Full-stack UX that keeps AI outputs usable and explainable.
About
Thoughtful systems, human rhythm.
ML/AI engineer
I’m Tanmay Dogra, a full-stack ML/AI engineer who specializes in building models, training pipelines, and the systems that ship them into real products. I focus on turning research ideas into working software, pairing clean data flows with reliable inference services and user-facing applications. I care about the engineering details that keep models stable in production, from monitoring to latency and scale. My goal is to make AI tools practical and trustworthy so teams can move from prototype to impact without losing momentum.
Focus
- Backend systems for ML (APIs, pipelines, model serving)
- Agentic AI workflows, RAG, and orchestration
- Production scaling, monitoring, and reliability
- Front-end product design for AI workflows
How I work
I move fast with clear structure, pair a design eye with engineering rigor, and ship experiences that feel grounded, calm, and intentional.
Featured case study
RAG Summarizer
A retrieval-augmented system that condenses long research documents into readable, structured summaries with source grounding.
RAG
Retrieval
Python
LLM Ops
More work
Additional highlights
Medical education
Meridien
A medical report reader for educational use that parses PDF medical reports, integrates Apple Watch exports, captures chronic illness histories, and answers patient questions about recent events. It holistically analyzes this data and provides planned suggestions to help patients understand their information.
Research topic
ANN vs. CNN: Feature Engineering vs. Deep Learning for Automated Coin Grading
A comparative study on the Saint-Gaudens Double Eagle gold coin, evaluating classic feature engineering + ANN baselines versus end-to-end CNNs for automated grading, with an emphasis on generalization and practical deployment tradeoffs.
Disclaimer: Meridien is for educational purposes only and does not provide medical advice. Always consult a qualified healthcare professional.
Contact
Let’s ship something that works.
If you’re building an ML product, a RAG/agent workflow, or taking a model from notebook → production, I can help you design the system, implement the backend, and make it reliable in the real world.
What I’m great at
Full‑stack for ML/AI workflows
- Agentic + retrieval systems (RAG, tool use, evals, guardrails).
- Backend + data pipelines (APIs, orchestration, queues/streaming).
- Production scaling (latency, cost, observability, reliability).
Include in your message
- Goal + users.
- Data shape (PDFs, text, time-series, images, etc.).
- Constraints (privacy, budget, latency, timeline).
Direct
Fastest ways to reach me
Email is best. I’ll reply with next steps, clarifying questions, and a realistic plan for shipping.
Typical response: 24–48 hours.