Forgetwell is built with a deliberately small stack: Go, Fiber, SQLite, and a vanilla-JavaScript frontend, shipped as a single binary. No frameworks for the sake of frameworks — speed and trust are the product. That system is now scaling: more users, more surfaces, and AI in the loop.
This is a builder role with real ownership. You propose a feature — or a product — prototype it in days, and carry it all the way to production: one person, end to end. Increasingly that also means building AI agents that automate workflows, inside the product and behind it: support, operations, testing, and the development loop itself.
What you will do
- Ship end to end: propose, prototype fast, then schema, API, UI, launch — one person carries an idea to production here
- Build AI agents that automate real workflows, in the product and in how we work
- Scale the system as usage grows: storage, sync, search, and serving — evolve the architecture to whatever the load demands
- Keep capture and recall fast — performance is a feature, not a metric
- Own quality: tests, migrations, observability, and calm operations
What we look for
- Strong fundamentals in at least one backend language (we use Go) and the willingness to work without heavy frontend frameworks
- Product sense: you notice what users need, propose it, and prototype in days, not quarters
- Experience building and operating large-scale distributed systems
- You sweat data integrity: migrations, backups, and edge cases
- You like small teams, short feedback loops, and owning problems whole
- Bonus: LLM inference and serving, CUDA and GPU experience, agent frameworks and MCP, SQLite internals, PWA or iOS