Shrikar Archak
What works in production, what doesn't, and the methodology to tell them apart. Battle-tested notes on fine-tuning small models, agent harnesses, and evals.
shrikar.comGENETICS AND GENOMICS
One epoch of contrastive fine-tuning lifts EmbeddingGemma from 0.73 to 0.82 NDCG@10 on insurance retrieval: a ten-minute, $0.65 GPU run. The full eval methodology, from raw HuggingFace pairs to trusted numbers.
GOOGLE
ACCOUNTING AND AUDITING
Zero false auto-approves and 92% accuracy. The useful part was discovering my measurement was broken three separate times, twice while every single test stayed green.
ACCOUNTING AND AUDITING
An agent that reads invoices, matches them against purchase orders, and decides. Exactly one step uses a model. Everything that makes it employable is deterministic code, and a test enforces that.
BANKING AND FINANCE
Mortgage files, insurance claims, and tax prep reduce to the same chase loop. I built an agentic AI workflow to run it: the agent owns the loop, a human in the loop owns the gates, and autonomy is earned gate by gate.
AI ETHICS AND GOVERNANCE
I built a tool-calling support agent from scratch plus a trajectory eval, then stacked five small reliability fixes and tested them on three models. What each fix buys depends entirely on how good the model already is.