/about/profile
Machine Profile
Grounded systems work across kernels, distributed telemetry, embedded tooling, and ML runtime infrastructure.
readme
I build software where state, failure modes, and performance characteristics matter. That usually means low-level systems, infrastructure with strong operational constraints, or tools that expose what the machine is actually doing instead of hiding it.
The through-line is not a language or a stack. It is a bias toward debuggability, strong interfaces, and evidence-based optimization. I care about the shape of logs, the quality of instrumentation, and the moment when an abstraction leaks under production load.
operating parameters
primary domains
kernels, infra, telemetry, edge systems, ML runtimes
preferred tools
Linux, perf, eBPF, GDB, QEMU, structured traces
engineering style
measure first, simplify interfaces, preserve escape hatches
failure posture
surface state early, design for triage and recovery
bio.log
[trace] built teaching kernels to understand scheduling and VM from the inside.
[trace] worked on telemetry paths where bandwidth and reliability fight each other.
[trace] shipped tooling for embedded and FPGA bring-up when generic tools were insufficient.
[trace] designed runtime controls for ML evaluation and observability pipelines.
current interests
filesystemsvirtual memoryscheduler designembedded observabilitytrace toolingGPU runtimesperformance forensics