Senior software engineer specializing in systems observability, security, and infrastructure for AI research.
Most recently Co-founder & CTO at Aethra Labs (YC W26), where I built an eBPF-based AI agent monitoring platform and telemetry analysis stack for frontier-model evaluations.
Previously led infrastructure for the Center for AI Safety's 256-GPU research cluster and WekaFS deployment supporting high-impact ML safety research.
Aethra Labs (YC W26) San Francisco & Berkeley, CA
Co-founder & CTO July 2025 – December 2025
Architected a system-level AI agent monitoring platform using eBPF (libbpf) to capture kernel-level traces (file I/O, network, process trees) with <3% CPU overhead, providing auditability independent of agent state.
Built a high-throughput telemetry ingestion and visualization stack (Python, TypeScript, Docker) that transformed raw system events into forensic-style replays for investigation of complex agent behavior.
Engineered AI-based trajectory monitors that automatically flagged suspicious or misaligned behavior and integrated them into real-time analysis pipelines for continuous safety evaluation.
Deployed privileged observability daemons inside Redwood Research evaluation environments, attaching to target PIDs to capture ground-truth runtime behavior for safety-critical model testing.
Executed a 24-hour rapid deployment of red-team infrastructure for xAI Grok evaluations, utilizing a LiteLLM proxy and OpenWebUI to enable structured interaction and centralized logging.
Raised $700,000 in seed funding from Y Combinator and prominent angels, including Geoff Ralston.
The Center for AI Safety (CAIS)
Lead Infrastructure Engineer February 2023 – July 2025
Owned infrastructure for a GPU cluster for ML research (OCI, Slurm) with 256 GPUs and a distributed parallel filesystem (WekaFS), supporting 125+ papers and 16,000+ citations; $3.5M+ annual budget.
Led deployment of converged 200TB WekaFS, reducing storage cost/GB by 90% for $900k+ in annual cost savings.
Mitigated EOL risks and enabled modern containerization and WekaFS support by executing a full cluster OS migration to Ubuntu 22.04, rewriting Ansible automation and migrating 200TB and 150M+ files.
Bidtellect (acquired by Simpli.fi)
Platform Software Engineer June 2020 – February 2023
Powered a platform processing 6B+ daily auction requests by building, debugging, and maintaining RESTful APIs, ETL pipelines, and SQL/NoSQL data systems.
Audited and optimized 700+ SQL queries, removed dead queries, and cut user login latency by 50% (6s to 3s) by improving a cache-loading query.
Supported team execution as Scrum Lead by scheduling ceremonies, unblocking engineers, and tracking sprint delivery.
Knoebel Institute for Healthy Aging
Research Assistant / Electrical Engineer September 2018 – June 2020
Designed and built an advanced neural recording apparatus (tetrode drive) improving reliability and performance compared to conventional electrodes.
Tested wireless neural recording devices and recorded rodent motor cortex activity to study the relationship between neural signals and motor movement for movement-prediction modeling.
Secured $5,500 in research grants.
Technical Skills
Systems & Security: Linux Kernel, eBPF (libbpf), System Instrumentation, Tamper-Resistant Telemetry, Perf Profiling
Aethra Labs (YC W26) San Francisco & Berkeley, CA
The Center for AI Safety (CAIS)
Bidtellect (acquired by Simpli.fi)
Knoebel Institute for Healthy Aging