Resume

Contact Information

Architect for control plane infrastructure team, guiding a team of more than ten engineers.
Supported regional expansion from ~10 to over 30 regions to satisfy customer demand.
Guided development of multi-cloud solutions to enable low-effort CI/CD across public, sovereign (national), and air-gapped clouds.
Implemented fully private/encrypted networking solution for customers using Azure’s Private Link / Endpoint mechanics.
Leading transition to IPv6+IPv4 support for all Azure ML services.
Designed solution for high-availability, multi-cluster Kubernetes backend, including dynamic nodepool scaling, atomic rollout+rollback.

Helped launch the general availability of Azure ML “v2” in late 2018, including addressing compliance, disaster recovery.
Designed and implemented CI/CD system for a 200+ engineer team to reliably ship updates to control plane microservices.
Guided efforts in porting Bing-internal experimentation / training services from .NET Framework (Windows-specific) to .NET Core + Docker container microservices.

Migrated all Bing source code from proprietary (Perforce-fork) source control to Azure DevOps Git.
Helped build and maintain packaging solutions (versioning, distribution) for C++ libraries shared across many repositories.

Designed, implemented, and shipped an internal metrics platform (similar to Prometheus) for Bing’s internal services, handling over 15,000 baremetal machines, each emitting over 1,000,000 metrics per minute.
Migrated Bing’s core query distribution and aggregation system from C++ to C# with no regression in performance (latency or throughput).
Designed and implemented a discovery mechanism for the ~100 distinct microservices, including dynamic graph building for dependencies, as well as optimistic caching.
Improved realtime query-serving network performance and reliability by implementing a custom load balancing and fault detecting mechanism for microservices hosted across >30,000 baremetal machines per region, across several regions.

Designed and implemented a service to provide dynamic timezone conversion utilizing user location data, and reactive to timezone database updates shipped as part of the core Windows infrastructure.
Implemented a system for realtime validation of “instant answer” services by reversing per-service grammars to ensure user results were returned as expected whenever data sets were updated.

Hands-on build out of Bing’s second-ever (and largest) datacenter, including scripted provisioning of Top-of-rack routers using serial connections and Python scripts.
Implemented a system to ship a large (100s of TB) dataset cross-country over a weekend in preference of physically shipping HDDs.

Current primary languages are Go, C#, Python
Experience implementing highly available, latency-sensitive distributed applications across multiple platforms + languages
Substantial experience with Windows, Linux, and BSD-based operating systems.
Familiar with many other languages, including C, C++, Bourne (again) shell (bash), PowerShell, Ruby, etc
Kubernetes, Docker, and containerization in general
Deep networking knowledge across many layers of protocols (IP + TCP/UDP, HTTP (1/2/3), BGP, Ethernet)
Extensive experience with many revision control systems, particularly Git, Perforce, and Subversion, with knowledge of Mercurial and CVS