I’m jabinwan, an engineer at a phone OEM, coding for 5+ years.
My path
- Systems kernel (first two years): familiar with the Linux kernel, C-focused kernel development for mobile devices;
- Mobile client (one year): Android Framework and System App work, adding the application-layer perspective;
- AI engineering (recent two years): transitioned to AI, currently focused on AI Agents — especially Agent Memory, observability, and evaluation.
Kernel → client → AI is a wide arc, but the core engineering instincts (observable, testable, rollbackable) carried all the way through.
What this blog covers
Notes on what I understand and what I’m learning:
- Design trade-offs in Agent Memory;
- Agent observability — knowing what an Agent did and why;
- Evaluation — measuring an Agent rigorously;
- Occasional throwbacks to systems and client-side days.
Contact
- GitHub: @jabinwan
- X: @jabinwan
- Email: jabinwan@gmail.com
Collaborate
Happy to talk about AI Agents (memory, observability, evaluation), mobile systems, or content; both email and GitHub are monitored.