AI Product · Consumer · Infrastructure · Google

Whitney Jenkins

Product manager with a background across infrastructure, consumer, and AI — from networking primitives at Google to products used by 100M+ people at Microsoft. I'm drawn to where AI meets everyday work: shipping tools and product surfaces that consumer and enterprise teams can adopt, trust, and grow with. I build end to end, including AI-native apps in my own time.

Let's Connect· PM · Google · NC · Product-led, AI-driven

I've spent my career working on products where the complexity has to disappear before the product can work. At Google that's networking infrastructure — the kind of plumbing that has to be invisible for AI workloads to scale. Before that it was OneDrive for 100M+ users and Azure Pipelines for millions of developers. The problem is always the same: find where something is harder than it needs to be, and fix that.

ROLEProduct Manager · Google
STATUSConsumer · Infrastructure · AI
LOCATIONNC, USA
EDUCATION
BS Information ScienceMinor Computer ScienceUNC Chapel Hill
ACCOLADES
🏆 2× Google Tech Impact Award WinnerFeatured Speaker @ Product School2019 Aspen Ideas Festival Scholar

Proven outcomes · Global scale

Shipped. Used. At scale.

Three features shipped in 2025, each one aimed at the same thing: removing a decision that developers shouldn't have to make in the first place.

My build loop

This is how I approach product development in the age of AI — synthesize the problem, ship something testable, then let real usage signal guide what's next.

01

Synthesis

PMF analysis, competitive signal, and user friction mapping. I use NotebookLM and Gemini to synthesize research fast — turning raw context into a clear problem statement before a single line of code.

// OUTPUT: PROBLEM_STATEMENT
02

Prototyping

From brief to testable build. Claude Code and Cursor handle the boilerplate while I stay focused on the core UX hypothesis. The goal is a working artifact — not a wireframe — in the hands of a real user.

// OUTPUT: WORKING_PROTOTYPE
03

Validation

Real signals replace assumptions. I validate early and often — with customers, stakeholders, and anyone close to the problem. Every iteration starts with evidence, not guesses.

// OUTPUT: VALIDATED_SIGNAL

This loop isn't theoretical — see it applied to a real build ↓

Shipped artifacts

GKE Accelerator Network Profiles

shipped · 2025
user friction

GPU/TPU workloads required manual configuration of multiple VPC networks and subnet interfaces.

outcome

Standardized GKE automated networking to provision all VPCs and subnets via a single-flag configuration.

impact

Removed manual setup for RDMA/GPUDirect connectivity across A3, A4, and TPU Trillium nodes. Available to all accelerator cluster customers on Google Cloud.

↗ View documentation

GKE Auto IPAM

shipped · 2025
user friction

Managing subnets required manual oversight and networking expertise to prevent scaling blocks.

outcome

Architected GKE Auto-IPAM to automatically handle subnet creation and range allocation.

impact

Removed the manual overhead of subnet management across GKE clusters, with proactive scaling to prevent deployment failures. Available to all GKE Standard and Autopilot customers.

↗ View documentation

GKE Multi-Subnet Support

shipped · 2025
user friction

Clusters were bound to initial subnet size; hitting limits required disruptive cluster recreation and migration.

outcome

Designed multi-subnet support, enabling clusters to assign additional subnets for new node pools.

impact

Removed the need for disruptive cluster recreation when hitting subnet limits, with up to 9x more address capacity available through additional subnets.

↗ View documentation

Professional evolution

How I got here

Consumer scale at Microsoft and GitHub. Platform depth at Azure and Google. AI velocity now. The through-line across all of it: make the complex thing disappear for the person using it.

  • Google
    Nov 2023 – Present

    Product Manager — Cloud Networking

    I own the networking strategy for GKE and GCE — the primitives that determine whether enterprise AI workloads can scale or stall. Every A3, A4, and TPU Trillium cluster on Google Cloud runs through features I shipped in 2025.
    SHIPPEDReduced the primary barriers to GKE adoption by architecting Auto-IPAM, Multi-Subnet support, and Accelerator Network Profiles. This transformed networking from a manual bottleneck into an automated utility, enabling enterprise customers to scale clusters beyond previous infrastructure limits.
    LEADERSHIPDrive feature and roadmap buy-in across engineering, design, and senior leadership. Coordinate cross-org to secure build commitment, aligning different priorities, timelines, and definitions of done behind one delivery plan.
  • Microsoft
    Feb 2021 – Aug 2021

    Product Manager II — OneDrive

    Consumer PM at scale — OneDrive served 100M+ active users across Business and Consumer. This is where I learned that consumer-grade product discipline — fast, obvious, delightful — is harder to maintain at platform scale than any technical constraint.
    PLATFORMDrove the unification of the OneDrive Business and Consumer backends — bringing two separate codebases onto a shared foundation, reducing duplicated engineering investment, and enabling a single shipping surface across platforms.
    SCALE100M+ active users across 190 countries. At that scale, a single UX decision affects millions of people's daily workflow — and the humility that requires shapes every product call I make.
  • GitHub
    Jul 2020 – Feb 2021

    Product Manager III

    Developer experience PM — Azure Pipelines is used by millions of engineers as a daily workflow tool. Consumer-grade expectations in an enterprise product: fast, obvious, and it works the first time. This shaped how I think about every developer-facing surface since.
    DELIVERYPrioritized Pipelines UX improvements targeting the highest-friction onboarding moments. The bet is that enterprise adoption is won or lost at initial setup. Teams that don't get to value in the first week rarely scale.
    IMPACTAzure Pipelines processes hundreds of millions of workflow runs. Shipping at that scale means thinking as carefully about edge cases and failure states as about the happy path.
  • Microsoft
    Apr 2017 – Jun 2020

    Program Manager — Azure DevOps

    Three years on Azure DevOps through its highest-growth period — owning product through the shift from VSTS to cloud-native CI/CD for enterprise teams at scale.
    SYSTEMSLed the rebranding and product roadmap for Azure Pipelines during the high-stakes transition from VSTS to Azure DevOps. Architected the global URL migration and core product experience, proving that successful infrastructure shifts depend more on user behavior and trust than technical specs.
    SCALEAzure DevOps and VSTS serve millions of developers globally. Every feature shipped went directly into the workflows of enterprise teams at scale.
  • Fidelity Investments
    Aug 2015 – Apr 2017

    Associate Systems Analyst

    Started in FinTech as a systems analyst and scrum master — where I learned that the PM fundamentals of specs, process, and stakeholder alignment apply at every scale.
    FOUNDATIONWrote feature specs that cut development time by 30%, drove agile process improvements that increased team velocity by 30%, and led a research study that expanded financial literacy training from 5 locations to nationwide — with 100% senior steering committee approval.
    EXPERIMENTATIONImplemented A/B testing on Genesys call routing technology — an early instinct toward evidence-based iteration that runs through every product decision since.

AI product lab

Building with AI, not just managing it.

Outside of work I build things — usually starting with a user problem and seeing how fast I can get to something testable. It keeps me honest about what the tools can actually do, and what it feels like to be on the other side of a product decision.

Product OS stack

COGNITIVE

// Research before building.

NotebookLMGeminiStitch
EXECUTION

// Agentic build.

Claude / Copilot / AI StudioCursor / Xcode / VSCodeGitHub
DEPLOYMENT

// Ship and scale.

Vercel

Prototype gallery

♟️
72h · Zero to TestFlight

GameNight Navigator

A board game helper built as a personal experiment in AI-native development. Brief to TestFlight in 72 hours using Claude Code, Gemini, and Xcode.
Tested with real users and iterated before the 72-hour mark.
Hover to expand
00hUser friction discovery and market-fit analysis via NotebookLM.
24hWorking prototype via Claude Code and Gemini. Architecture validated.
72hLive on TestFlight. Real users. Real signal.
StackSwiftUISwiftDataClaude CodeGeminiXcode
🌐
Prompt-to-Prod · Live

whitj.com

This site. A self-evolving AI-native portfolio — orchestrated as a single-file React app, iterated via Claude Code agentic loops.
In production on Vercel — refined through the same agentic loop, not a static handoff.
Hover to expand
ArchitectureSingle-file React. No build pipeline friction.
Iteration loopClaude Code agentic sessions replace a design-dev handoff.
ConstraintSingle file, no backend. Forces prioritization — if it can't be done in JSX it doesn't ship.
StackReactTailwindClaude CodeVercel
＋
In build now
A consumer AI app. Early signal promising. Back before Q3.

Writing & speaking

Publications & talks

Three Google Cloud Blog posts and a KubeCon NA 2025 talk — writing and speaking as part of shipping, not separate from it.

⚡ talk↗
KubeCon NA 2025
Kubernetes IP Management: From Core Concepts to Strategic Solutions
co-presented with Ivy Zhuang, Google
blog↗
Google Cloud Blog · Aug 2025
Announcing multi-subnet support for more scalable GKE clusters
with Ivy Zhuang, SWE
blog↗
Google Cloud Blog · Aug 2025
IP address management made easy: Announcing auto IPAM for GKE clusters
with Gina Yeh, SWE
blog↗
Google Cloud Blog · Dec 2025
Unlocking GKE's Full Potential: The Flat Network Decoded
with Abdel Sghiouar, Sr. Cloud Dev Advocate

Product OS

How I think

Four ideas. One operating model.

The distance between idea and working product has shrunk

With the right tools, a PM can go from problem statement to something testable in hours rather than sprints. I've done it — brief to TestFlight in 72 hours using Claude Code, Gemini, and Xcode. I think this changes what good product process looks like, and I'm still working out what that means.

Something testable beats something polished

A working app in someone's hands — even rough — tends to surface things that mockups don't. The constraint of shipping something real forces clarity about what the product actually needs to be. I'd rather learn that early than after three weeks of refinement.

Infrastructure decisions are product decisions

What you expose as configurable, what you abstract away, and what you make automatic shapes what every team building on top can do. Working on GKE networking has given me a lot of respect for that — the choices made at the primitive level ripple outward in ways that are hard to undo.

Consumer and infrastructure aren't opposites

Many PM careers go deep in one direction. I've had the chance to work across both — consumer products at Microsoft and GitHub, platform infrastructure at Google. I think the combination is useful, especially for AI products that have to be both reliable and easy to use. Still figuring out how to make the most of it.

Let's build something that matters.

I'm always interested in connecting with people working on hard problems — whether that's infrastructure, AI, or consumer products at scale. If something I've built or written resonates, feel free to reach out.

Let's Connect · Quickest on LinkedIn or email