Resume

The complete record – where I’ve worked and what I actually did there. If you just want the shape of it, About has the three-sentence version.

NVIDIA (December 2024 – present)

Senior AI Infrastructure Engineer, DGX Cloud Operational Excellence (June 2026 – present)

Revamping DGX Cloud’s incident management practice around on-call health and learning from incidents.

Senior AI Infrastructure Engineer, DGX Cloud Infrastructure Security (January 2026 – June 2026)

Founding member of the team, focused on building the data pipeline for agentic security workflows.

Senior AI Infrastructure Engineer, DGX Cloud COSI – Colos, On-prem, and Systems Images (December 2024 – January 2026)

Repaved one-off GPU clusters with thousands of H100 GPUs that weren’t deployed with infrastructure-as-code into end-to-end IaC-managed clusters.
Identified and automated the response to Linux, Slurm, and GPU (XID) errors as well as other hardware failures.

Staff Platform Engineer, Platform Core (May 2022 – December 2024)

Moved Retriever – Honeycomb’s query and storage engine – from EC2 to Kubernetes, which first meant finding and fixing a Linux kernel bug that crashed nodes during LVM snapshots.
Tracked a performance regression down to the Go gRPC library doing an excessive number of DNS lookups on Kubernetes; fixed it by deploying NodeLocal DNSCache.
Added a safety check to Kafka rolling restarts: a ZooKeeper lock that won’t let a broker upgrade until every replica is back in sync. Only one broker gets to be down at a time, no matter who pushes the button.
Cut Kafka producer latency by 20% with garbage collection and heap tuning.

Senior Software Engineer, OS/Compute (April 2021 – November 2021)

Built the service that emits metrics and events whenever the OOM killer fires on any of 250,000+ EC2 nodes, then worked with the affected teams to right-size their instances and stop the unexpected restarts.
One of four engineers on call for the Base OS team – the Linux layer under every instance and container at Netflix.

Senior DevOps Engineer, Personalization Infrastructure (February 2019 – April 2021)

Ran the platform for the ~60 engineers and researchers training and deploying the ML models behind Netflix recommendations – row selection, box art selection, billboard selection.
Operated Apache Spark clusters on Mesos across tens of thousands of EC2 instances.
Helped develop and deploy a method of sharding Mesos behind ZooKeeper after we hit its scaling ceiling around 12,000 nodes.
Built the system that tracked data fidelity and model quality, so regressions got caught – and root-caused – before members noticed them.

Co-VP, Trans* Employee Resource Group (February 2019 – November 2021)

Led the Education and Talent Acquisition pods, then served as Co-VP of the ERG itself.
Developed and delivered training with the Inclusion & Diversity team on how to be a better colleague to trans and gender non-conforming people.
Consulted with content teams on trans representation, both in front of and behind the camera. (How that ended is its own story.)

Senior Systems Engineer, Big Data (February 2018 – January 2019)

Planned and executed a zero-downtime migration of the production Kafka clusters – 20+ billion events a day – from a commercial distribution to Apache Kafka, dropping the licensing costs on the way out.
Deployed Kafka on Kubernetes with StatefulSets, securely mirroring production data to support moving the Global Data Platform into the public cloud.
Replaced a commercial monitoring vendor with Puppet, Telegraf, Jolokia, and Grafana – $100K+ a year cheaper, and we could finally see what was actually going on.
Debugged open source and in-house applications in Scala, Java, and Python to help developers speed up their jobs and stop wasting resources.

UNIX Systems Engineer (January 2015 – February 2018)

Built and ran the telemetry pipeline behind Overwatch, World of Warcraft, and Hearthstone – service health for 40+ million monthly players.
Designed and deployed the globally distributed ZooKeeper architecture that gave Overwatch service discovery and picked the game server with the best latency for everyone in the match.
Migrated a multi-petabyte Hadoop data warehouse to new hardware with zero downtime by dual-writing through Kafka – 4x the compute and memory, enough to actually run Spark and Impala.
Moved the global fan-in data pipeline from RabbitMQ to Kafka: 3x the capacity, trillions of events retained, and the end of losing data.
Cut KPI report time from 3 hours to 10 minutes by migrating jobs from Streaming MapReduce to SQL on Hive, Spark, and Impala.

Solutions Consultant

Deployed, configured, and tuned dozens of Apache Hadoop clusters running Accumulo and HBase – kernel, network, filesystem, and memory management.
Owned the Hadoop and Active Directory/Kerberos integration, eliminating a vendor dependency.

IT Manager, then Systems Engineer

Ran corporate IT through the spinoff to 42six.
Deployed the Hadoop and Accumulo clusters the developers built on, kept the public-facing web servers up, and stood up the Jenkins CI that built everything.

IT Manager / Network Administrator

Ran a New Jersey K-12 district’s network – 700+ computers and a redundant gigabit fiber backbone installed during a $20M renovation.
Laptops for every teacher on a zero-growth budget, and an internship program for students considering a career in IT.