Systems engineer focused on ML infrastructure, healthcare tooling, and developer productivity.
I write Rust for performance-critical paths, Python for pipelines, and ship things that work in production.
batch_forge — Bare-metal inference engine for JAX/Equinox models in Rust. Zero-copy SafeTensors loading, hand-optimized Metal and Vulkan compute kernels for transformer workloads.
Provectus — Regulatory document drafting prototype (Swift). Generates CTD/CSR DOCX with sentence-level citations, source traceability, and exportable audit reports.
primer-scout — Fast Rust CLI for primer off-target scanning on FASTA files. Handles mismatch tolerance and reverse-complement search.
EvidentenceBinderV2 — Structured evidence and signal management for clinical/decision workflows.
graphrag-prototype — Graph-based retrieval augmented generation experiments.
DOCENT — TypeScript tooling project (active).
- tracel-ai/burn — Contributed to the Rust deep learning framework (14.9k stars). Focus area: performance and backend portability.
- rustfs/rustfs — Contributed to the S3-compatible high-performance object storage system (25.6k stars).
Languages — Rust, Python, TypeScript, Swift, Julia, SQL, R
Cloud — AWS, Azure, GCP
Tooling — Docker, Airflow, dbt, Linux
- Optimize for real-world constraints: reliability, latency, cost, security
- Prefer explicit interfaces, measurable pipelines, honest documentation
- Build in public to make the work reusable
Co-author — Big Data for Smart Energy, CRC Press, 2021