Divanshu Sharma — Applied AI / ML Systems Engineer

run 00 · cached

“whoami”

▸ intent: about
▸ tool: get_about()
▸ synthesis: cached

DIV-1 online. It answers for Divanshu Sharma — a founding engineer who ships reliable AI products. His focus is LLM evaluation, inference and reliability. Right now he owns Uniiq's AI admissions platform — including a decision-planning layer that cut critical-constraint violations from 70% to 0% in offline evaluation; before that, privacy-preserving ML research at Yale and three years of quant research at WorldQuant BRAIN. The depth is public: 24 merged PRs to AI infrastructure like Mastra and EleutherAI's lm-evaluation-harness, and 47 from-scratch rebuilds of the modern AI stack. Each of the 62 stars behind this card is one of his real systems — hover one, or ask anything and watch the answer assemble.

Divanshu Sharma

Applied AI / ML Systems EngineerLLM evaluation, inference & reliability

Founding Engineer @ Uniiq · Mumbai, India

Open to AI engineering & founding roles

Founding engineer who ships reliable AI products. At Uniiq he owns the AI admissions platform: a stateful intake hardened so model outages can't corrupt student state, 40+ critical vulnerabilities resolved, and a Decision Twin that cut critical-constraint violations from 70% to 0% in offline evaluation. Before that: privacy-preserving deep learning research at Yale and three years of quantitative research at WorldQuant BRAIN.

40+
critical vulnerabilities resolved
Uniiq · prompt-injection & data-leak risks
24
merged upstream PRs
Mastra · EleutherAI lm-eval · +18 bugs fixed upstream
88.08%
ChestMNIST accuracy under MPC
from-scratch DL library, Yale
70% → 0%
critical-constraint violations
Uniiq Decision Twin · offline evaluation