Railtrack OD on DMP Di1
Embedded dual-model railway obstruction pipeline running BiSeNetV2 and YOLOv8s on the DMP Di1/iCatch A3000 NPU with live camera streaming and integration into a fleet DMS/ODS dashboard.
I build AI systems and embedded ML pipelines — from a $40 SoM running a 5 fps obstruction detector to production Python APIs. End-to-end, shipped.
I'm a software engineer who builds things end-to-end — from C++ inference pipelines running on edge AI hardware to Python CLI tools to React frontends. I get bored building in one layer and genuinely curious about how everything connects.
Over the past few years I've shipped railway obstruction detection systems on embedded NPUs, built AI agent frameworks, designed an Obsidian-native productivity system, and created this very portfolio tool to track it all. The thread is: I care about production. Not demos.
For freelance clients: I take projects from idea to deployed product. I'm direct about tradeoffs, fast to iterate, and I write documentation people actually use.