I'm Hank Sha.

|

Currently seeking Applied AI / FDE roles for Summer 2026. Reach out →

AI engineer building production LLM systems at the intersection of technical depth and customer impact.

Skill Map

Production Systems

Python TypeScript Node.js Docker PostgreSQL MongoDB Git Vercel n8n

ML / Research

PyTorch HuggingFace LangChain Scikit-learn Claude Code

Data + Analysis

R Python Scikit-learn XGBoost

Building/Learning in Public

I write to clarify my own thinking.

Selected Writing

More on Medium →

Projects

Data4Good Hackathon

1st Place

UCSB 2026. Won the Civic Engagement & Policy track at the Data4Good hackathon.

View on Devpost →

Biotech Intelligence Engine

Personal Project

Multi-source biotech intelligence platform — aggregates SEC filings, clinical trials, FDA calendars, and news. Built to surface investable signals before they hit mainstream feeds.

AI Job Applier — Chrome Extension

20min → 3min Personal Project

Built a Chrome extension that scrapes job descriptions and auto-fills application forms across major ATS platforms (Greenhouse, Lever, Workday). Reduced per-application time from ~20 min to under 3 min. Handles dynamic DOM injection, content-script messaging, and multi-step form traversal.

Exploring How LLMs Improve Graph Quality Using Topological Signatures

Research

Research investigating how large language models can enhance graph quality through topological signature analysis in tasks like node classification and graph unlearning.

Activation Steering with a Feedback Controller with Emphasis with Application in Cross-Layer Behavior and Integral-Windup Tests

Research AI Safety

Activation Steering with a Feedback Controller with Emphasis with Application in Cross-Layer Behavior and Integral-Windup Tests

Selected Work

Applications of the above. Each follows: Context → Action → Lessons