Apoorva
Verma
Software engineer shipping full-stack products, plus the retrieval systems and evaluation harnesses that keep applied AI honest.
The chat below is one of them. Every answer is grounded in my real work before it reaches you.
Experience
Software Engineer · Oracle
2023 — nowBackend services and applied-AI work — RAG and LLM evaluation for enterprise search.
SWE Intern · Microsoft Engage
2022Built an ML-driven recommendation feature end-to-end.
SWE Intern · J.P. Morgan
2021Data tooling for an internal risk-analytics platform.
Fellow · MLH
2021Open-source fellowship — shipped to a production codebase.
How I evaluate retrieval
From rag-eval-harness: a reproducible comparison over 15 papers, scored with DeepEval plus a custom citation-faithfulness metric. Semantic chunking with e5-large wins on context precision; the real lever is retrieval, not the model.
Featured projects
rag-eval-harness
RAG over 15 arXiv papers with exact page and section citations. The real project is the eval harness: a reproducible retrieval and chunk×embedding comparison scored with DeepEval and a custom citation-faithfulness metric.
FounderSignal
Market-intelligence platform that mines and validates early startup signals. Full-stack Next.js app with data pipelines, scoring, and a dashboard.
promptguard
Regression testing for LLM apps: pytest for prompts. Write checks against prompt outputs so quality regressions fail in CI instead of reaching users.
Notes on grounding & eval
View all writing →Engineer who measures what others guess
I'm a software engineer at Oracle. I ship full-stack products end to end, and I go deep on applied AI: the retrieval and evaluation work that keeps LLM systems correct in production. Python and TypeScript, backend to frontend, with the unglamorous testing most people skip.
Before that, a path through some places that taught me to ship: Microsoft Engage, J.P. Morgan, and Major League Hacking. I care about reproducible benchmarks, calibrated judges, and answers that cite their sources.