// hello, world

I'm Sherry

Senior AI engineer shipping production LLM systems and voice AI. I build agents that hold up on live traffic, ship open-source, and write about what I learn along the way.

Voice agents handling 10% of live calls at sub-800 ms
Book an intro callView projectsResume

// projects

Things I've built

Open-source tools and experiments. Each one started with a problem I couldn't stop thinking about.

Insurance Voice AI

Production

Orchestrator routing live inbound callers to 4 specialist agents (health, car, travel, bike). Handles 10% of live inbound calls at sub-800 ms average latency.

Voice AIMulti-AgentLLMReal-time

claude-audit

Active

Command-line code auditor. Fast static checks, plus an optional audit where Claude investigates the code with read-only tools.

TypeScriptAI AgentsSecurityCLI
1.5k+15npm

llm-ocr

Active

Convert PDFs and images to clean Markdown with vision LLMs (Gemini, OpenAI, Claude, Ollama). Python library and CLI.

PythonLLMVisionPDF
3k+3PyPI

async-limiter

Maintained

Asyncio rate limiter for Python that enforces a concurrency limit and a requests-per-period limit together.

PythonAsyncioRate Limiting
15k+PyPI

titans-unofficial

Maintained

Unofficial PyTorch implementation of the paper "Titans: Learning to Memorize at Test Time".

PythonPyTorchResearchDeep Learning
8k+PyPI

WebLM

Active

Web-to-Markdown API with an open-source Python SDK that extracts typed structured output into Pydantic models.

PythonFastAPIPydanticLLM
1k+PyPI
View all projects →

// now

What I'm up to

Last updated:

Building

Wrapping up two years of shipping production voice AI: a multi-agent system that handles live insurance calls at sub-800 ms latency, plus the full-stack work around it. Now contributing to open source, maintaining my own packages, and building new things I can't stop thinking about.

Thinking about

What it takes to run AI agents on live traffic, not just in demos: latency budgets, evaluation, and knowing when to hand off to a human.

Learning

End-to-end product pipelines and scalability. How real systems go from idea to production, and how they hold up when the traffic actually arrives.

// contact

Let's build something together

Whether you have a project in mind, want to collaborate on open-source, or just want to talk about AI, I'm always up for a good conversation.