Meadow.OS — a .NET operating system for microcontrollers
Wilderness Labs · 2020 – Present
Problem
Wilderness Labs set out to let developers build secure, connected
device firmware in C# and .NET — on microcontrollers with a few
dozen megabytes of RAM and flash, a class of hardware normally
programmed in C.
What I did
Led OS engineering for
Meadow.
I ported the Mono runtime to STM32 hardware on the NuttX RTOS,
integrated mbedTLS for device TLS, designed the full-system
over-the-air update machinery — from custom bootloader to MQTT
delivery — co-designed the .NET low-level I/O API (GPIO, SPI,
I²C, interrupts), ported TensorFlow Lite for Microcontrollers
with .NET bindings, and ran the OS release process.
Outcome
A production embedded operating system shipping on commercial
hardware: developers write, monitor, and remotely update device
applications entirely in .NET.
Mono — runtime engineering for cross-platform .NET
Xamarin & Microsoft · 2014 – 2020
Problem
Mono
was the open-source .NET runtime underneath every Xamarin iOS and
Android app. Garbage-collector behavior, runtime reliability, and
the release pipeline behind a >1 GB macOS SDK needed
sustained core-team work.
What I did
Improved the SGen generational garbage collector, hardened the
runtime's native crash-reporting machinery, ported file and
network I/O subsystems from Microsoft's CoreFX, maintained
bockbuild
— the build and packaging system behind the Mono macOS SDK — and
integrated runtime releases into the Xamarin products.
Outcome
Runtime and tooling releases shipped continuously to Xamarin's
developer base, through the Microsoft acquisition and beyond.
Seurat — finding cancer mutations in genome pairs
TGen · 2009 – 2014
Problem
Detecting somatic mutations means comparing a patient's tumor and
normal genomes at whole-genome scale — with early-generation
sequencing noise, as much a statistics problem as a software one.
What I did
Designed a Bayesian method for tumor/normal comparison and wrote
Seurat,
its open-source implementation
(first-author paper, BMC Genomics).
Helped design and operate TGen's HPC cancer-genomics pipelines, and
contributed to the pilot phase of the
1000 Genomes Project.
Outcome
Co-author on 41+ peer-reviewed papers cited over 11,000 times; the
1000 Genomes dataset became the reference map of human genetic
variation for hundreds of subsequent studies.