Jason Vallery · Longmont, Colorado

AI storage architect and infrastructure product leader

I have spent most of my career on storage and distributed systems. From 2017 through 2025, I led Azure-side storage work supporting OpenAI and Microsoft AI. My job was to keep training data and checkpoints moving without leaving expensive GPUs waiting on storage. After Microsoft, I spent part of 2025 and 2026 at VAST Data working on the data architecture and economics behind training and inference.

I work from the data path outward: placement, routing, capacity, clients, failure, recovery, and the operating loop that keeps the design tied to production. I turn those constraints into decisions that engineering teams, customers, partners, and executives can use.

[email protected] · vallery.net · linkedin.com/in/jvallery

Areas of depth

What I know well

Professional experience

Storage and systems, from support through architecture

About thirty years of work, with more detail on the systems that explain what I do now.

Oct 2025–2026

VAST Data

VAST Data

Vice President, Cloud Product Management

I worked on the cloud data architecture and economics behind AI training and inference.

  • Model how concurrency, context length, throughput, utilization, and write patterns move data across GPU memory, host memory, flash, network, and persistent storage.
  • Connect KV-cache capacity and write load to SSD endurance, latency, system design, and total cost.
  • Turn those constraints into product direction, commercial models, partner work, and field guidance across hyperscalers, neoclouds, and AI-infrastructure customers.
Sources

Aug 2017–Oct 2025

Microsoft

Microsoft

Azure Storage product and architecture leadership

I led Azure-side storage architecture, product, capacity, and operations work supporting OpenAI and Microsoft AI.

  • Designed a checkpoint path that used memory and local or peer NVMe for speed, with Azure Blob as the durable recovery layer.
  • Helped develop Scaled Accounts, which presented one namespace across multiple storage clusters and required placement, routing, and rebalancing as capacity grew.
  • Led Azure Blob’s storage and data workstream for Project Fairwater, including regional placement, pre-staging, service-side data movement, capacity, performance, and reliability.
  • Owned the customer-facing Blob experience and an operating loop built around capacity reviews, telemetry, regressions, major incidents, root-cause analysis, and roadmap changes.
  • These were team outcomes. My role was infrastructure and product leadership underneath the models, not model research. OpenAI named me under Microsoft Azure Support in its 2019 Dota 2 report.
Sources

Apr 2013–Aug 2017

Microsoft

Microsoft

SharePoint Premier Field Engineer → Cloud Solution Architect

I moved from SharePoint field engineering into Microsoft’s early Azure field organization.

  • Designed cloud architectures, built proofs of concept, and helped customers and software partners move from on-premises products to Azure services.
  • Served as a technical escalation point and translated architecture choices into migration and business decisions.
  • Authored a five-file Azure Resource Manager quickstart for a high-IOPS VM and storage pool; the pull request and README follow-up both merged.
Sources

Jul 2012–Apr 2013

Neudesic

Neudesic

Pulse product management

I worked on product management for Pulse, Neudesic’s enterprise-social product. My formal public title was Director, Technical Solution Sales.

  • Connected customer and field requirements to product positioning, technical sales, and adoption guidance.
Sources

May 2010–Jul 2012

Avanade

Avanade

Manager → Group Manager

I led SharePoint and Office 365 delivery for enterprise customers.

  • Directly managed distributed onshore and offshore engineers and project managers.
  • Owned delivery quality, staffing, escalation, customer communication, and technical direction.
Sources

Oct 2008–May 2010

NewsGator Technologies

NewsGator Technologies

Technical Project Manager → Solution Architect

I led enterprise RSS and social-collaboration delivery for large SharePoint environments.

  • Moved from project management into solution architecture as the work became more technical.
Sources

2001–Oct 2008

SHL Group

SHL Group

Support, software, solutions, and integration delivery

I progressed from client support into software development, solutions, project management, and technical integration delivery.

  • Built and delivered web recruitment and assessment systems and led HR-XML and SOAP integrations.
Sources

1996–2001

Earlier work

Earlier work

IBM Global Services · University of Colorado

I began working in IT at fourteen and moved from university systems administration into enterprise support.

  • Administered AIX, Windows NT, and Novell systems and built Perl and JavaScript tools for the University of Colorado Office of Financial Aid.
  • Provided desktop and application support for Lucent Technologies while working for IBM Global Services.
Sources

Independent work

What I build outside the job

These projects put architecture, code, operations, and failure in the same loop.

2024–present

Verdify

I build an AI-assisted greenhouse control loop with my adult son. Deterministic ESP32 firmware owns safety; AI can tune only bounded parameters, and telemetry verifies the result.

See the work

2024–present

Personal agent fleet

I operate a private pod-per-repository fleet of coding agents on Kubernetes using GitOps, isolated worktrees, model routing, observability, and explicit controls.

See the work

2001–2016

Earlier products and projects

I founded Brio Solutions, ran NewsPicker and Mine Feed, and turned a Gmail contact-extraction utility into MineMyMail. Each began as a problem I wanted to solve for myself.

See the work

Education

No completed degree

I attended Front Range Community College for roughly three semesters while working full-time. I do not have a completed college degree. Most of my technical education came through work, public troubleshooting, independent projects, and continued hands-on practice.

More context

Work, talks, and background

A resume is useful, but it leaves out why the work mattered to me. The About page tells that story; Work goes deeper on the systems; Speaking collects the talks and interviews.