About
Ann and George
build Latent Signal.
We are software engineers who spent over two decades each building desktop, web, and mobile software. We left our corporate careers to build independently — learning, tinkering, and shipping on our own terms.
George started his career building high-performance trading software for day traders in C++ and MFC at a startup. From there he moved to medical image management systems for the web (C#, .NET, ASP WebForms) with PACS, and then spent over a decade at Intuit - first as a Distinguished Engineer on financial transaction understanding in QuickBooks Accounting, then as Director of Engineering for the QuickBooks Financing as a Service team.
After 28 years of building software across desktops, web, and enterprise systems, George left Intuit in January 2025. He now spends his time learning ML fundamentals, experimenting with LLM APIs and coding agents, and building projects with Ann. He is particularly interested in where AI tools actually hold up under real use and where they fall apart.
George blogs occasionally at george.chiramattel.com/blog.
Ann has built software across ultrasound imaging at Philips Healthcare (C++), application platforms at SAP (Java), desktop tax software at TurboTax (.NET, C#, C++), and mobile apps at TurboTax and QuickBooks Self-Employed (iOS, Swift, Kotlin). At Intuit, she served as Director of Engineering for QuickBooks Mobile, then led QuickBooks Platform — the database, API, and platform layer of the QBO monolith. Her last role there was leading the Generative AI initiative in QuickBooks.
Ann left Intuit in July 2024 after 14 years. She now builds apps that use LLMs in meaningful ways (KeepSeek for household item tracking, Beast Mode for workout tracking), learns ML math through Math Academy and MIT OCW, and experiments with every coding agent she can get her hands on - Claude Code, Codex, Gemini CLI - to understand what AI-assisted development looks like in practice.
Ann blogs regularly at annjose.com/post and plays piano when she is not coding.
Our stance
What we believe
We believe that understanding fundamentals matters more than knowing the latest framework. The tools change every year. The principles underneath - how systems work, why they fail, what makes software hold up - change much slower. That understanding is what lets you evaluate new tools honestly, debug problems you've never seen before, and know when a shortcut is safe to take.
We believe AI makes experienced developers faster, and that this is a good thing. We use coding agents and LLMs every day. We also read every line of generated code before it ships, because the tool does not know what it does not know - and neither do we, unless we look.
We believe the best way to understand a technology is to build something real with it. Tutorials and blog posts help, but the gaps in your understanding only show up when you're solving an actual problem. That's why we build apps, enter hackathons, and write about what we find - including the parts that didn't work.
What we've built
Projects
HN Companion
hncompanion.comA browser extension that helps you read Hacker News faster by summarizing discussions into meaningful chunks.
Live product
Email Memory
An app that synthesizes emails across multiple people and timelines to build meaningful context - for example, consolidating all communications with a contractor and city about installing a heat pump.
Hackathon winner
Periscope
A tool that visualizes how the context window of a coding agent grows during a session and suggests when to fork or rewind.
Hackathon finalist
Get in touch
Say hello.
Questions, ideas, or just want to connect - reach out.