Skip to content
View gbfish's full-sized avatar
2 followers · 1 following

Achievements

Achievement: Arctic Code Vault Contributor

Achievements

Achievement: Arctic Code Vault Contributor

Block or report gbfish

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
gbfish/README.md

Illustration: walking from mobile apps toward local LLM agents, MCP, and point-in-time data, notebook in hand

Software engineer, 10+ years. Mobile (React Native, Swift), Rust, embedded, and lately: making AI agents actually work.

What I'm doing now

  • Running local LLMs through real agent loops and writing down where they break — agent-lab. Every conclusion in that repo is traceable to a raw trajectory; if a number isn't backed by a run, it isn't in the README.
  • Researching MCP as infrastructure: what a server can and can't do on iOS, why MCP gateways became a category, and where the protocol has no answer yet (physical-safety envelopes, RBAC).
  • Quantitative trading as a data-engineering problem: point-in-time datasets, cost-aware backtests, and the honest arithmetic of small accounts.

How I work

I write things down. Research notes, dead ends, and corrections stay in the repo next to the code. If I was wrong, the correction is marked, not silently edited.

Elsewhere

Writing: gbfish.github.io (RSS) · Houston, TX · equalbean@gmail.com

Pinned Loading

  1. agent-lab agent-lab Public

    Experiment bench: where does local model + Goose break when running a real agent loop? Every conclusion traceable to raw trajectories.

    Python