ML engineer and GenAI consultant helping enterprise teams ship AI products that work. Currently, Principal Consultant for a Tech Consulting company, building production RAG systems, autonomous agents, and LLM infrastructure for energy and mining clients. Currently, helping one of largest Mining companies in the world build GenAI solutions. Previously at Dataiku, Coursera, and Udacity.
I spend most of my time at the intersection of GenAI product management and hands-on ML engineering - scoping what to build, then building it. Retrieval pipeline architecture, LLM routing, evaluation frameworks, and on-prem inference for clients with real constraints.
On the side, I build small products:
| Product | What it does |
|---|---|
| Inquiro | AI-native notebook first coding agent |
| FieldNotes | Autonomous agents mine the world's open data, surface the stories hidden inside, and publish every step of their reasoning. |
| DocsForLLM.dev | Crawls documentation sites and composes optimized llms.txt files for LLM context windows |
| DoINeedPermits.com | Tells homeowners exactly which permits they need before starting renovation work |
I write about what I learn at sshtomar.github.io. Some recent posts:
| Date | Post |
|---|---|
| Feb 2025 | RAG for Miners |
| Jan 2025 | The Engineer Who Writes |
Before GenAI, I spent years in the learning and skills space - user research at Udacity, skills gap analysis for Fortune 500s at Coursera, and AI/ML workflows for banking clients at Dataiku across Southeast Asia.
In 2017, I was a Research Fellow at the University of Chicago's Data Science for Social Good program, working with Rayid Ghani and the World Economic Forum to identify illegal fishing vessels using satellite imagery and AIS data.
I also partnered with Cambridge University and FourthRev to architect data science curriculum for working professionals.
I've contributed to tools across ML infrastructure, data science education, and developer workflows.
- claude-code-skills-social-science -- Claude Code skills for rigorous social science research: DID, RCTs, regression diagnostics, and more
- llm-txt -- Generate llms.txt from documentation sites



