Hi, I’m Srinidhi Ramanujam. I build production AI systems for regulated enterprises. I work on the parts that come after a promising demo: architecture, data integration, evaluation checks, controls, rollout, and user trust.
I have spent 17+ years shipping ML and AI systems, including two granted US patents from production ML work. Today my focus is enterprise GenAI for BFSI workflows, including financial analytics agents, text-to-SQL, agentic ingestion, and the evaluation and governance work that makes those systems usable.
Selected public work
- Lessons From a Year of Shipping LLM Analytics Products for Enterprises: Six lessons from building enterprise analytics products that use large language models.
- Context engineering experiments: Controlled experiments on long context, retrieval, structured prompts, latency, and answer quality.
- Quant Magic: Natural language to SQL financial analytics over public S&P 500 data.
- Patents: US12069122B2 and US11310162B2.
Explore the site
- Writing: Long-form essays on production AI, evals, data systems, and engineering work.
- About: A concise professional bio and contact links.
- Fiction: Where I indulge in a bit of story writing.
- Links: Annotated bookmarks, prototypes, and notes worth saving.
Spotlight
Spotlight
Lessons From a Year of Shipping LLM Analytics Products for Enterprises
Six lessons from shipping LLM analytics products for enterprises: start with workflows, fix the data model, define evaluation, layer security, make answers inspectable, and plan for model and vendor change.