Notes from the GitHub Trenches: What I’ve Been Working On Lately

It’s been a busy coding season for me, with a few projects evolving in parallel. I thought I’d break from the usual narrative and give a quick tour of what’s been happening in my GitHub lately — essentially, a peek into the code trenches:

• Traider: I’ve been refining the trading logic and integrating a real-time data feed. One recent commit was a complete overhaul of the strategy evaluator; now Traider can pull in live market data and simulate trades on the fly. I also squashed a nasty bug that was causing memory bloat during backtesting. It’s running smoother now, and inching closer to handling a live demo without falling over.

• Coach: A lot of under-the-hood improvements here. I implemented the OpenAI-to-Ollama fallback system, so Coach won’t go silent on me if the cloud AI fails (that was a fun challenge, essentially adding a backup brain). I also spent time debugging a WebSocket connection issue that was preventing Coach’s scheduled reminders from firing. After a few late nights and one “aha!” moment, the issue is resolved — now Coach reliably sends my morning SMS and syncs with my Notion notes every day.

• Build-TMS: This project is in its early stages but progressing. I drafted the JSON schema for Transfer Modules and built a basic interface to create and edit these modules. The app can now take a batch of my notes, vectorize them, and produce a .tm file (the “brainfile” format) which I can load into an AI agent. It’s still rough around the edges, but seeing the first end-to-end test (from input text to a working module) was super motivating.

Those are the big three at the moment. Of course, there are always smaller experiments and one-off scripts in my repo — a few data visualizations here, a bot there — but Traider, Coach, and Build-TMS have been my main focus. It’s fun (and sometimes a little chaotic) juggling multiple projects, but I love how each one teaches me something different. And writing it all down like this helps me appreciate the progress that might not be obvious day-to-day when I’m deep in the code.