Divanshu Sharma

Founding engineer who ships reliable AI products — currently leading the technical turnaround of Uniiq's AWS-deployed AI advising platform: 40+ critical vulnerabilities resolved, LLM sanitization and recoverable AI failures, LCP cut from 3.3s to 0.7s. Before that: privacy-preserving deep learning research at Yale and three years of quantitative research at WorldQuant BRAIN.

1. Founding Engineer — Uniiq

   May 2026 — present

   • Led the technical turnaround of inherited legacy code, resolving 40+ critical vulnerabilities and adding prompt and response sanitization against injection and data-leak risks.
   • Reworked student onboarding into a three-phase AI conversation with follow-up questions, persisted transcripts, structured validation, rate limits, and recoverable AI-failure handling.
   • Implemented role-aware profile-completion enforcement and international phone capture across protected APIs, sign-up, setup, and profile flows.
   • Expanded institution and opportunity administration with deep-partial updates, nested Mongo normalization, validation, and external-ID uniqueness safeguards; added 1,199 lines of automated tests across 12 modules.
   • Raised the web performance score from 73 to 93, cut LCP from 3.3s to 0.7s, and reduced INP to 130ms after code-splitting 21 routes and optimizing delivery.

2. ML Researcher — Yale University

   Oct 2025 — Mar 2026

   • 88.08% ChestMNIST test accuracy in 5 epochs at batch size 512 — trained under MPC.
   • No external autograd: layers, backprop and optimizers derived by hand.
   • CNN stack proposed upstream to 0xTCG/sequre across four pull requests.

3. Research Consultant — WorldQuant BRAIN

   Nov 2022 — Jul 2025

   • Developed and backtested 15+ alpha signals; average Sharpe ratio 1.8.
   • GOLD LEVEL (Top 1%) and Stage 1 Qualifier, Global Alphathon 2022.

• Uniiq — AI Student-Advising PlatformFounding engineering · Full-stack AI product · 2026

  40+ critical vulnerabilities — resolved in inherited legacy code, including prompt-injection and data-leak risks; 73 → 93 performance score — LCP cut from 3.3s to 0.7s; INP reduced to 130ms.

• Privacy-Preserving Deep Learning via MPCML research · Yale University · 2025–26

  88.08% test accuracy — ChestMNIST, strided-convolution CNN, 5 epochs @ batch 512; 0 external autograd deps — backprop, conv & pooling layers derived and implemented by hand.

• Indic ASR Optimization & Consensus EvaluationSpeech · Evaluation research · 2025

  >48% absolute WER reduction — FLEURS Hindi benchmark vs. baseline Whisper-small; 177,000 unique terms — normalized via Devanagari engine + reverse transliteration.

• RenAIssance OCR — Historical Document PipelineComputer vision · NLP · 2026

  Word accuracy 97.96% → 98.17% — full corpus, after LLM post-processing; CER 3.44% → 0.00% — on the held-out 183-page validation set.

• Paged KV-Cache Engine (vLLM-style)LLM inference systems · 2026

  Copy-on-write forking — ref-counted blocks shared across forked sequences; Radix-tree prefix cache — shared prompt prefixes served without recompute.

• Graph Data Explorer AIApplied LLM product · 2026

  765 nodes / 877 edges — typed knowledge graph from 19 source tables; Two-stage LLM query engine — NL → traversal → NL, fully sandboxed.

• Deep Learning for CMS Detector PhysicsScientific ML · ML4SCI / CERN CMS · 2024

  80% AUC — electron-vs-photon classification, custom ResNet-15; −30% training time — GCN matching the GAT baseline for momentum regression.

• Mastra (mastra-ai/mastra) 11 merged PRs — 29 PRs and 18 bug reports to the TypeScript agent framework (Aug–Sep 2026). 11 merged directly — codemod and CLI fixes for its migration tooling — and all 18 bugs he found and patched are fixed upstream, one merged with him as co-author. #58 of 466 contributors by commits.
• EleutherAI lm-evaluation-harness 4 merged PRs — 25 PRs to the standard LLM evaluation harness (Aug–Sep 2026): 4 merged — CLI argument parsing, request-cache robustness, and a restored benchmark task — and 21 in review.
• Experiential (experientiallabs) 2 merged PRs — 20 PRs (Sep 2026): 2 merged model-provider fixes — paginated Anthropic model discovery and Gemini page-token handling — and 18 in review.
• Sequre (0xTCG) proposal, unmerged — 4 PRs bringing the Yale work upstream — CNN layers and an MPC ChestMNIST training pipeline (#41 alone: +3.5k lines across 54 files). Not merged yet; #42 is open.
• p5.js Web Editor (Processing Foundation) 4 merged PRs — 10 PRs (2023–24), 4 merged: a keyboard shortcut for adding files, the Korean translation of keyboard shortcuts, and two UI fixes.
• OpenCode (anomalyco/opencode) proposal, unmerged — 22 fix PRs to the open-source AI coding agent (210k★), Aug–Sep 2026 — 21 open, awaiting maintainer review.
• Also merged: vulnerablecode, Music Blocks, PyNN 3 merged PRs — A license-URL fix to aboutcode-org/vulnerablecode, a close-button fix to Sugar Labs' Music Blocks, and a docs-layout fix to NeuralEnsemble's PyNN.

Languages

Python (primary) · C++ · TypeScript / JavaScript · Rust · SQL · Bash · Java

Modeling & training

PyTorch · TensorFlow · Keras · scikit-learn · Hugging Face Transformers · PyTorch Geometric · CNNs / GNNs · Whisper · DeBERTa / RoBERTa

LLM systems

Gemini · structured AI outputs · RAG · SFT · LangChain / LangGraph · Model Context Protocol · prompt engineering · Qdrant · Chroma · Ollama

Serving & MLOps

FastAPI · Docker · ONNX · GitHub Actions CI/CD · MLflow · Airflow · AWS (SageMaker, EC2) · GCP · Azure · Streamlit

Data

NumPy · Pandas · Polars · CatBoost · PostgreSQL · MySQL · MongoDB · SQLite

Web & product

React · Next.js · Node.js / Express · Django / Flask · Tailwind CSS · server-sent events · system design

B.E. Computer Science — University of MumbaiCGPA 8/10 · 2021 — Jun 2025

Machine Learning Specialization (DeepLearning.AI, 2023) · Deep Learning Specialization (DeepLearning.AI, 2023) · TensorFlow Developer Certificate (DeepLearning.AI, 2024) · GANs Specialization (DeepLearning.AI, 2024) · Meta Front-End Developer (Coursera, 2023) · Postman API Fundamentals Student Expert (Postman, 2023)

• GOLD LEVEL (Top 1%) & Stage 1 Qualifier — WorldQuant BRAIN Global Alphathon 2022
• Kaggle 2× Expert
• LeetCode contest rating 1,821 (top 7.5%) · CodeChef 3★ (1662) · 4,400+ problems solved
• GitHub: Pull Shark ×3 · Starstruck ×2 · YOLO · Quickdraw
• 195 public repositories · 337 stars