Open to new roles · Now: Frontier AI Team, Microsoft Business Operations

I build AI agents and automations that do the work your team shouldn't have to.

Copilot agents, Foundry agents, AI-infused power automation workflows, and intelligent dashboards; built hands-on, from discovery to production, as Customer Zero for Microsoft's own business processes. Backed by 20+ years shipping for Microsoft, Dell, and AWS.

Work With Me See What I Build

20+

years in tech delivery

80%

process time cut with AI workflows

136

Microsoft Learn badges · 24 trophies

30

LinkedIn recommendations

Current Role · Dec 2025 - Present

Senior Power Platform Strategy & Automation Consultant - Microsoft Business Operations, Frontier AI Team

Agentifying and automating Microsoft's business processes as Customer Zero: partnering directly with business teams, then building, shipping and measuring the agents, apps and automations myself with Power Platform, Copilot Studio, Microsoft Foundry, Dataverse, Microsoft Graph, Entra ID, GitHub Copilot and Azure AI.

What I Build

AI that ships, not slideware.

Every engagement ends with something running in production: an agent answering questions, a workflow moving data, a dashboard driving decisions.

🤖

Copilot & Custom AI Agents

Agents built in Copilot Studio, Microsoft Foundry and Azure OpenAI, secured with Entra ID and grounded in Microsoft Graph, that answer questions, enforce compliance, generate plans, and take action across your Microsoft 365 estate. I built a Copilot agent at Microsoft that automated feature and release planning with 16 criteria-based inputs.

Copilot StudioMicrosoft FoundryAzure OpenAIMicrosoft GraphEntra IDAgentic AI

⚡

AI-Infused Automation

Workflows that trigger on schedules or events, transform and route data between systems, send alerts, verify permissions, and eliminate tedious work - with AI in the loop where it adds judgment.

Power AutomateAzure Logic AppsFrontierScout

📊

Intelligent Dashboards & BI

Power BI and Fabric dashboards that aggregate critical data at a glance - with Copilot-assisted analytics on top. Don't have the data yet? I'll build the system that collects, organizes, and surfaces it.

Power BIFabricCopilot in Power BIKPI Automation

🚀

AI-Era Release & Program Management

End-to-end delivery for complex programs: Azure DevOps administration, release planning, sprint analytics, and the change management that gets AI adopted - not just installed.

Azure DevOpsRelease PlanningAgile/LeanChange Mgmt

🧩

Full-Stack AI Building

Low-code where it is fast, code where it is not: Power Apps and Dataverse front ends, custom connectors, APIs and services in JavaScript/TypeScript, Python and C#, built with GitHub Copilot and GitHub CLI, while I still own the architecture, quality and reliability.

Power AppsDataverseMicrosoft GraphGitHub CopilotTypeScript / Python / C#

🏗️

Platform Engineering & Migration

SharePoint Online architecture, cloud migrations, and secure collaboration platforms - the durable foundation your AI and automation layer runs on. 500,000+ assets migrated at 99.9% uptime.

SharePoint OnlineM365Entra IDCloud MigrationGovernance

How I Work

From ambiguous problem to production, owned end to end.

I don't hand off after discovery or design. I partner with the business, build the solution myself, and measure whether it changed how work gets done.

01

Discover

Sit with stakeholders, map the real workflow, challenge assumptions and agree on success measures.

02

Prototype

Use AI-assisted development to test feasibility and show value early, with a path to production.

03

Build

Agents, flows, apps and integrations, grounded in business data with human-in-the-loop controls.

04

Launch

Deploy, support-ready, adopt and iterate, with security, DLP, testing and monitoring in place.

05

Measure

Track adoption, cycle time, manual effort avoided and quality, then reuse what worked.

💡

Reimagine, don't just automate

I look for work to simplify, redesign, agentify or eliminate, not only the existing steps to speed up.

🛡️

Responsible by design

Security, privacy, responsible AI, access controls and lifecycle management are built in from the first prototype.

♻️

Built for reuse

Reusable agents, components and templates, such as the shared preference layer behind the Theo family, that raise what the whole team can build.

Spotlight · Sept 2026 launch

MBO Frontier AI Agent ROI Dashboard.

From black box to clear returns: I took a spaghetti data model fed by fragmented sources, rebuilt it as a clean star schema on Dataverse, and put a lightweight Power BI report with HTML5 KPI cards on top of it.

MBO Frontier AI Agents / Assets dashboard overview page in Power BI — The new MBO Frontier AI Agents / Assets dashboard

Microsoft Business Operations · Frontier AI Team

Repair the legacy model. Reconcile the content. Make AI impact easy to understand.

The original agent ROI report had grown large, bloated and hard to trust: fragile DAX with agent-specific branches, many-to-many joins and bidirectional filters that made lineage hard to reason about. Leadership needed one number they could believe for how much capacity AI agents hand back to the business.

I treated it as a data problem, not a UI problem: repaired the old SharePoint model to establish a trustworthy baseline, migrated to a Dataverse-only star schema, used AI to compare and explain every difference between old and new totals, then rebuilt the report as a fast, lightweight front door.

10→2

fragmented sources consolidated into one fact + one dimension

5

report views: Overview, ROI, Details, Treasury, KPI explainer

12

slide launch walkthrough I built and presented

01

Understand

Decode the old system: broken DAX, many-to-many calendar joins and bidirectional filtering.

02

Repair

Fix the legacy SharePoint model first so the old numbers became an accurate baseline.

03

Rebuild

Move to Dataverse and align a clean star schema: a Log fact table and an Agent dimension.

04

Reconcile

AI-assisted compare, explain and true-up: what changed, why, and which differences are expected.

05

Explain

HTML5 flip cards so every KPI shows its purpose, source and formula on the reverse.

Before

Ten SharePoint-era sources stitched together
Fragile calculations with flagged field references and per-agent branches
Many-to-many joins and bidirectional filters muddying the totals
Heavy, slow pages and numbers nobody could trace

After

Dataverse-only star schema, documented table by table
Actual impact separated from estimated capacity
Fiscal-year context and drill-through from summary to the supporting log
A dedicated Treasury view, plus submit-an-agent, library and feedback actions

From spaghetti to a star.

One grain, one direction of filtering, one place to define each measure. The core is a two-table Dataverse model, an activity Log fact and an Agent dimension, filtered one way by a fiscal calendar, so every total on the page traces back to individual runs.

Power BI (.pbip)DataverseStar schemaDAXPower Query MHTML5 / CSS cardsSharePoint OnlineAI-assisted reconciliation

The launch deck

Scroll sideways, or click any slide to step through all twelve.

01 · A new foundation, a clearer view

02 · The old experience needed more than a facelift

03 · Untangle the schema before trusting the totals

04 · First, repair the old SharePoint model

05 · Build on Dataverse and align the final model

06 · AI helped explain differences, not just rewrite code

07 · A lightweight, fast-loading front door

08 · Separate actual impact from estimated capacity

09 · Follow the summary back to activity

10 · A focused Treasury view

11 · Every KPI explains its purpose, source and formula

12 · Less report bloat, more clarity

Featured · 2026

The Theo agent family.

Five production agents built on Power Automate, Dataverse and Azure DevOps. Each one solves a specific operations problem, and they share a common preference and notification layer.

Theo DQ V2.1 preview

Data-quality advisor

Theo DQ V2.1

Wakes up at 3 AM Pacific, finds ADO items whose case and engagement status disagree, scores a fix with a confidence level, and asks the PM to approve with one click.

Power AutomateAzure DevOpsAI Builder

Read the case study →

Theo V3.1 preview

Preference-aware notifications

Theo V3.1

“Ask Once. Remember Forever.” A reusable framework that remembers whether each person wants Teams or Outlook, and routes every alert accordingly.

DataverseTeams cardsChild flows

Read the case study →

Theo Engagement Manager preview

Daily engagement scoring

Theo Engagement Manager

Scores every active engagement daily, flags what is at risk, and feeds an executive dashboard with the month’s tracking status.

Power AppsDataverseCopilot prompts

Read the case study →

Theo ADO Query Manager preview

Self-service alert builder

Theo ADO Query Manager

A hosted visual WIQL builder with Entra sign-in. Anyone can subscribe to an Azure DevOps query, pick recipients and a send time, and manage it under My Alerts.

Entra IDWIQLFlaskDataverse

Read the case study →

Frontier AI fleet preview

Five agents, one operating model

Frontier AI fleet

Five purpose-built agents deployed, live and compounding in value, with a portfolio rhythm that keeps intake, priorities and blockers visible.

Agent opsADO analyticsPortfolio rhythm

Read the case study →

Case Studies

Seven recent projects that moved the needle.

The problem, what I built, and what changed.
Click any screenshot to enlarge.

01Theo agent family

Theo ADO Query Manager

Turning “please set up an alert for my ADO query” from a developer task into a five-minute self-service flow.

The problem: Program managers wanted proactive Azure DevOps email alerts, but each one meant writing WIQL, wiring a flow and hoping the schedule was right.
What I built: A hosted, guided wizard: connect to an ADO org and project, choose or build a query (paste a link, pick a saved query, build visually, or enter raw WIQL), pick recipients, format and cadence, then confirm. Sign-in uses Entra ID. Confirmed subscriptions land in a Dataverse table that a Power Automate dispatcher reads. A “My Alerts” screen lets people edit cadence or cancel. Light and dark themes.
The outcome: Alerts are now self-service: per-person emails or a digest, user-selected send times and stop conditions, with no developer in the loop.

FlaskEntra IDWIQLPower AutomateDataverse

Query Manager recipients step — Step 3 · choose who gets the email

Query Manager dark theme query step — Dark theme · pick a query three ways

Query Manager cadence step — Cadence and stop conditions

02Theo agent family

Theo DQ: the data-quality agent

An agent that fixes drifting ADO data while people sleep, but still lets a human make the call.

The problem: Case state and engagement status in Azure DevOps kept falling out of sync (cases closed while the engagement was still in progress), and cleanup was manual and reactive.
What I built: V1 runs every morning at 03:00 Pacific, queries ADO for mismatches, writes status back and keeps an audit trail. V2.1 adds an AI advisor: it looks up the rules matrix, evaluates each case, assigns a confidence score and writes a recommendation. The PM receives one email per person with a proposed change and a Yes/No button that updates ADO directly.
The outcome: Daily, rule-based hygiene with human-in-the-loop approval, per-item confidence and a full audit trail, instead of periodic clean-up sprints.

Power AutomateAzure DevOpsAI BuilderOutlook actionable messages

Theo DQ V2.1 AI advisor design — V2.1 · the AI advisor

Theo DQ V1 workflow — V1 · the nightly workflow

Theo DQ rules matrix — The rules matrix the advisor reasons over

03Theo agent family

Ask once. Remember forever.

A reusable preferences framework so every notification arrives where the person actually wants it.

The problem: Different people prefer Teams or Outlook, daily or monthly digests, and nobody wants to be asked the same question twice.
What I built: Three reusable child flows (get, set, delete preference) backed by Dataverse. The first notification offers the choice in an adaptive card; the answer is saved and every later message routes automatically. V3.1 packages this as a six-flow production backbone with modular flows and error isolation, so one failure never stops the rest.
The outcome: One framework reused across agents, with self-service preference management and permission-aware delivery.

DataverseTeams adaptive cardsChild flowsOutlook

User preferences framework one-pager — The preferences framework at a glance

Theo V3.1 production backbone

Notification and preference confirmation cards — Step 1 and 2 of the user experience

04Theo agent family

Engagement intelligence and the Frontier AI fleet

Daily scoring of every engagement, rolled up into one executive view across five agents.

The problem: Leaders had no real-time view of engagement health. Status was reconstructed monthly from spreadsheets, and at-risk items were found late.
What I built: Theo Engagement Manager reads ADO engagement data daily, scores every item with an AI-generated tracking score, writes results to Dataverse and surfaces them in a Power Apps dashboard, with proactive emails to owners of items falling below threshold. The Frontier AI fleet view ties the five agents together with a shared operating rhythm: intake, prioritize, in flight, blocked, delivered.
The outcome: A single source of truth for engagement health, refreshed daily without manual effort.

Power AppsDataverseCopilot promptsADO analytics

Theo Engagement Manager overview — Executive dashboard overview

Engagement intelligence daily scoring — Daily AI scoring at scale

Frontier AI fleet overview — The fleet: five purpose-built agents

05Analytics

Rebuilding enterprise AI analytics on a trusted data foundation

From fragmented spreadsheets to a governed Dataverse model and a 15-page Power BI report.

The problem: Agent and asset reporting depended on seven workbooks and shared models that could not be trusted or reconciled.
What I built: Migrated reporting from SharePoint and Excel to Dataverse, built crosswalks and dimensions, rewired Power Query and reconciled historical and production sources. Rebuilt the CCE Continuous Improvement report across 15 pages and 152 visuals on ADO OData, with dynamic KPI narratives, MoM comparisons and RAG measures.
The outcome: A defensible ground truth, with 11 of 12 production measures matched across fiscal months and defects (duplicate IDs, double-counting, fiscal sorting) resolved.

84449

records in one trusted source

152

visuals rebuilt

15

report pages

47

ADO epics integrated

Power BIDAXPower QueryDataverseADO OData

MBO Frontier AI dashboard · a lightweight, fast-loading front door

ROI & Impact · actual impact vs. estimated capacity

Agent / Asset Details · follow the summary back to activity

06Microsoft Cloud

Success Hub Release Plan Agent

A Copilot Studio agent that creates a feature release plan through conversation and routes it for approval.

The problem: Creating a release plan meant filling a long ADO form, and incomplete or inconsistent entries slowed review.
What I built: A Copilot Studio agent that asks for 16 criteria-based inputs conversationally (product area, feature type, dates, scope, metadata), creates the ADO feature, then triggers Teams approvals for review, documentation and publishing, with comments written back to the work item.
The outcome: Faster, more consistent plans and a traceable approval trail, helping the Release Readiness workflow cut processing time by 80%.

Copilot StudioPower AutomateTeams approvalsAzure DevOps

Release plan agent conversation — Conversational intake

Teams approval request

Release plan flow architecture — Flow architecture

07Microsoft Learning

Intranet platform and ticketing automation

The foundations of the automation work: a portal platform, a ticketing system and a cloud migration.

The problem: A 20+ person team served 52 business groups and 5,000+ requests a year through manual intake.
What I built: Directed web development for the Microsoft //Learning intranet, designed a SharePoint template system, and engineered an automated ticketing system on SharePoint, ADO and Power Automate: five forms feeding three flows each, covering approvals, notifications and sign-off.
The outcome: 50% lower cost and 60% faster response times, plus 500,000+ assets migrated at 99.9% uptime.

SharePointSharePoint DesignerInfoPathPower AutomateADOProgram management

Automated ticketing system process — End-to-end production process

SharePoint template system

Learning portal home — The Learning portal

Side Quests - After Hours Lab

Projects, client work, family tools, experiments, and things I couldn’t leave alone.

AI agents, live sites, PWAs and automation pipelines, designed, built and shipped end to end.

Client build · Gates Painting · Live in Teams

Agent Katie: an AI office manager that reads, remembers, emails and texts.

An Azure-built AI office manager for a commercial painting company, live in Teams. Inside: what she does, eight capability tiles, Twilio SMS and the full Azure stack.

Gates Painting is a commercial painting company and a Microsoft vendor. Microsoft asked them to show how they're adopting automation and AI, and I built their answer inside their own Microsoft 365 tenant: a knowledge agent that captures how the office actually runs built to serve office staff, field crews and customers from the company's real mailboxes, calendars, files and job data.

Built on Microsoft Foundry with retrieval over Azure AI Search and Azure OpenAI, wired to Microsoft Graph, and deployed to Microsoft Teams, where it's live and answering.

Twilio SMS

Two-way texting, live Nov 2026

Staff text a toll-free number and Agent Katie answers from the same knowledge base
Fully automatic outbound replies, locked to an allowlist of approved numbers
Consent captured through a web opt-in form instead of paper
Started on Azure Communication Services, moved to Twilio for toll-free messaging

Mailbox RAG

Indexes the main inbox plus five shared mailboxes (info, media, billing and more) with app-only Graph auth.

Three-tier knowledge

Office and HR folders, the public website as JSON, and a Job Board snapshot, unified into one Foundry knowledge base.

Live calendar

An indexed calendar export plus a real-time getCalendarEvents tool: a Logic App on Graph exposed to the agent as OpenAPI.

Sends email, safely

A send_email tool from her own mailbox, with a confirm-recipient-and-draft step before anything goes out.

Email auto-replies

Email the agent and get an answer back. Allowlisted senders only.

Katie Triggers

Qualifying emails (new leads, POs, billing invites) post to a Teams alerts channel and log to a SharePoint list.

Leadership dashboard

A Power BI view over the tracker list so the owner can prioritize leads and act on new work.

Next up

WorkMax time-tracking data, a custom neural voice, and specialist agents for change orders, scheduling and accounting.

Azure & Microsoft 365

Microsoft Foundry Agent ServiceAzure OpenAIAzure AI SearchAzure Logic AppsMicrosoft Graph (app-only)Microsoft TeamsExchange OnlineSharePoint OnlineEntra ID app registrations

Power Platform

Power AutomatePower BISharePoint lists

Beyond Microsoft

Twilio Programmable MessagingSupabaseOpenAPI toolsPythonWorkMax API

AA

Account-Ant Family Budget App

Single-file PWA · private by default · beta

A private, single-file PWA that turns a bank CSV into a pay-period-aware budget with an opt-in Claude assistant. Inside: how it works, the seven dashboard tabs, the AI features and 13 app screenshots and launch ads.

“Know exactly what you can spend today without blowing your budget.” A personal cash-flow dashboard that turns a bank CSV export into a pay-period-aware budget. It detects income, recurring bills and spending categories, then shows what's safely spendable now, what each week looks like and where the month will end. All data stays in the browser.

How it works

Import a Chase (or any Date/Description/Amount) CSV
Auto-detect income, bills, categories and subscriptions in 1–2 seconds
Setup wizard confirms pay schedule, housing and weekly budgets
Enter your balance to drive Spendable Now and the month-end estimate

Seven dashboard tabs

Overview: bill + payday calendar with running balances
Spending and Savings: trends, over-budget alerts, savings-goal calculator
Fixed and Cash Left: recurring obligations and week-by-week budgets
Transactions and an AI-powered Tax Planner

Ask AccountAnt (AI)

Opt-in, bring-your-own Anthropic API key
Browser calls Claude directly; no middleware server
Sends numbers only, never identity
“Can I afford a $500 purchase?” · “Where am I overspending?”

My Budget AccountAnt · import a bank CSV to get started

Overview · weekly bill and payday calendar with Spendable Now

This Week · discretionary budget pills

Fixed · every recurring obligation in one place

Ask AccountAnt · AI answers from your real numbers

Launch creative · Know exactly what you can spend today

Feature tour

Stop guessing, start knowing

Ad concept

Welcome to the future of finance

Excel is officially dead

Your money. Your rules. Your privacy.

Feature highlights

PWAJavaScriptService worker (offline)Claude API.ics calendar exportlocalStorageNetlify

Private beta

AT

Multi-skill algorithmic trading pipeline

Agent chain · Claude skills

Five Claude-skill agents chained from research to execution, with a risk layer that can veto any trade. Inside: the agent chain, the live JG-CowboyTrader strategy and its trade journal.

Five specialized agents that hand off to each other as a chain. Each one has a single job, and the risk layer can stop a trade before it's ever built.

AnalyzeSentiment and market-regime research

→

BrainSignal scoring and position sizing

→

RiskManCircuit breaker and exposure veto

→

Order ExecutionBuilds the order ticket

→

Trade AutopsyAI post-mortems and rule-change proposals

Live as JG-CowboyTrader: Donchian-channel breakouts with ATR-sized units, pyramiding into winners and trailing 2N stops across a seven-symbol universe, running on a paper account, with every trade logged to a journal dashboard.

JG-CowboyTrader live status — JG-CowboyTrader · Turtle-style breakouts on a paper account

Trade journal dashboard — Trade Journal · P&L, win rate, calendar and drawdown

Agent orchestrationClaude skillsRisk controlsFeedback loopPaper trading

Private build

Chilly Wind Revival · home

About the band

Moments on the road and booking

+2 more

CW

Chilly Wind Revival

Band website · custom domain

A site for a Colorado jam band, built so the band can edit and publish it themselves without ever touching Netlify.

Local editor package for content changes
Publishes through a Power Automate flow I own (HTTP trigger → Netlify deploy)
Hosted on the band's own domain

NetlifyPower AutomateNo-access publishing

Visit the site →

Open roles ranked by fit

JS

Job Search Agent & Dashboard

Scheduled agents · private artifact

A private hosted dashboard that tracks scanned roles, fit-scores each one against my skills, and queues one-click applications.

Daily job scan across major boards and career sites
Hourly application worker processing the queue
Fit scoring, filters and status tracking in one view

Scheduled tasksAI fit-scoringClaude artifacts

Private build

RiffQuest · note highway, chord pads and tab view

RQ

RiffQuest

Guitar trainer PWA · real-time pitch detection

A guitar chord and riff trainer that listens through the mic and scores what you play.

Scrolling note highway with six string lanes matching the guitar
Live chord diagrams with a dimmed "next chord" preview
Watch, Practice and Perform modes; speed, loop and streaks
PDF drop folder for importing new songs; installable on Android

PWAWeb Audio APIPitch detection

Play RiffQuest →

Command Center · one board runs the whole jobsite

Executive roll-up

Daily operations view

Campus crew map and assignments

Office worker view

Crew phone view

Comms view

+6 more

GP

Gates Painting Operations Dashboard

The demo that started the engagement

A role-based operations dashboard for a commercial painting company, built to show what their data could look like.

Executive roll-up with On track / Watch / Needs action status
15+ views: Monthly, Daily, Foreman, Campus Crew, Field Office, Planning, Products, Comms
"Watch These First" trends panel, dark mode, single data block

HTML5JavaScriptNetlify

Private build

All sweaters with price history

Side-by-side comparison matrix

Store-to-store price and purity comparison

Historical trajectory and re-sale verification

Buy window and starred drawer

+4 more

AC

The Cashmere Drawer

Price chaser · family tool

Tracks cashmere sweaters across 11 retailers and flags when a favorite drops in price.

Price-drop radar with priority alerts and a 90-day-low view
Compare drawer for 2 to 6 items and a personal watchlist
Buy-window report; sort by value score, filter by store, fiber, color and size

Price trackingValue scoringNetlify

Family tool

Today · workout, meals and hydration

Weekly rhythm

How to do every move

Eat well, feel good

Grocery list by store section

+4 more

AW

Wellness Planner

Family tool

A weekly fitness, meal and grocery planner in one place.

Short daily workouts with tap-through exercise instructions
Weekly menu, daily-plate guide, swaps and snack ideas
Grocery list grouped by store section that resets every Monday

HTML5Mobile-firstNetlify

Family tool

X

x-social-response-agent

Social listening · human-approved posting

Monitors X for Power Automate and Power Platform conversations, then helps me show up with something useful to say.

Drafts replies and scores each one before review
Approval dashboard: nothing posts without a yes
Posts approved replies automatically

AI draftingX APIApproval workflow

Private build

PG

PaintGauge

Computer vision · React

Snap a photo of a 5-gallon bucket and Claude vision estimates how much paint is left.

Animated SVG fill gauge for the estimate
Persistent gallery of past submissions

ReactClaude visionSVG animation

Prototype

TV

LG Smart TV OneDrive Movie Viewer

Custom webOS app

A custom webOS app that browses and plays my own movie library straight from OneDrive on an LG smart TV.

OneDrive sign-in and streaming playback on the TV
Companion ThumbnailGridMaker (Python + ffmpeg) embeds a 2×2 preview grid into each MP4

webOSMicrosoft GraphPythonffmpeg

Personal build

Proof

Measured results, not promises.

A sample of outcomes from AI and automation work at Microsoft, Dell, and AWS.

80%

Processing time cut by consolidating 56 work items into one AI-powered "Release Readiness" workflow at Microsoft Cloud.

16

Criteria-based inputs in a Copilot agent that automated feature & release plan creation, accelerating planning cycles.

25%

Sprint planning time reduced via custom ADO templates and real-time story point tracking.

60%

Faster response times from an automated ticketing system built on SharePoint, ADO, and Power Automate - at 50% lower cost.

100%

On-time delivery across Microsoft Industry Cloud releases - Manufacturing, Healthcare, Sustainability, Sovereignty, Retail.

99.9%

Uptime while migrating 500,000+ assets, documents, and sites from on-prem SharePoint to the cloud.

View the full impact portfolio →

Experience

Two decades of delivery. The last five, AI-first.

Senior Power Platform Strategy & Automation Consultant - Microsoft Business Operations

Frontier AI Team · Dec 2025 - Present

Mission: agentify and automate Microsoft's business processes, putting the latest AI capabilities to work inside Microsoft as Customer Zero, in direct partnership with business teams.
Own solutions end to end: discovery, prototype, build, production launch, adoption and iteration, using Power Apps, Power Automate, Copilot Studio, Dataverse, Microsoft 365 extensibility, GitHub Copilot, GitHub CLI and Azure AI.
Build agents and AI-powered workflows grounded in business data, connected to tools and actions, with human-in-the-loop controls. See the Theo agent family.
Extend low-code with full-stack engineering (JavaScript/TypeScript, Python, C#, Microsoft Graph, custom connectors and APIs) when low-code alone is not enough.
Define success with stakeholders and measure outcomes: adoption, cycle-time reduction, manual effort avoided and quality, while building reusable agents, templates and patterns for the wider team.

Release Manager / Automations Engineer - Microsoft Cloud

via Sonata Software · 2022 - May 2025

Created a Copilot agent automating feature and release plan creation with 16 criteria-based inputs, accelerating planning cycles and stakeholder alignment.
Authored Power Automate workflows consolidating 56 work items into a single "Release Readiness" item - 80% faster processing and better cross-team visibility.
Orchestrated releases for Microsoft Industry Cloud SaaS across Manufacturing, Healthcare, Sustainability, Sovereignty, and Retail - 100% on-time, private preview to GA.
Built Power BI dashboards integrated with ADO to monitor execution, bugs, readiness, and change management, reducing defect rates.

Release Manager / Business Intelligence Developer - Dell Digital

2021 - 2022

Delivered performance metrics and insights driving executive decisions for a 1,000+ member organization.
Automated organization and normalization of thousands of Epics and Features for PI planning; introduced auto-refreshing Power BI dashboards adopted across business units.

Technical Program Manager - Microsoft Learning

via Denny Mountain, Allyis Inc. · 2013 - 2020

Directed web development for Microsoft's //Learning intranet - 20+ member team, 52 business groups, 5,000+ annual requests.
Engineered an automated ticketing system (SharePoint + ADO + Power Automate) cutting costs 50% and response times 60%; migrated 500,000+ assets to the cloud at 99.9% uptime.

Earlier roles - AWS, Windows Azure, and more

Website Manager - AWS.Amazon.com

2012 - 2013

Directed AWS website strategy; homepage and navigation improvements drove a 25% increase in sign-ups and conversions; led responsive templates project improving cross-device UX by 30%.

Website Manager - WindowsAzure.com

via Denny Mountain Media · 2008 - 2012

Ran Windows Azure's website plus 14 localized sites; platform migration cut hosting costs 40% and improved performance 50%; global traffic management delivered 99.99% availability.

Technical Project Manager - Impresys

via Denny Mountain Media · 2006 - 2008

Led technical projects for Microsoft engagements including the Azure Services platform launch at PDC; delivered demos, sales readiness materials, and BI resource kits on time and above 20% margin.

Earlier: Flash Developer (Filter Talent) · Project Manager (Pixel-Craft) · Creative Services Manager (FastSigns)

2000 - 2006

Interactive demos for Microsoft Office 2007 adoption, e-commerce builds, branding, and design for clients including Real Networks, the Seattle Mariners, and Tesoro Petroleum.

Credentials

136 Microsoft Learn badges · 24 trophies - AI-heavy and current.

Verified on Microsoft Learn. More than a third cover AI, agents and machine learning, including the newest on Microsoft Foundry, MCP and multi-agent orchestration, alongside the Power Platform stack that surrounds them.

47

AI, Agents & Machine LearningMicrosoft Foundry, Agent Framework, MCP agents, Work IQ & Web IQ, RAG, speech & language, Copilot stack, responsible AI

27

Power BI & AnalyticsReport design, Fabric semantic models, end-to-end analytics, Power Query

17

Power AppsAdvanced canvas apps, Dataverse, app source code

14

Power Platform GovernanceSecurity, environments, Center of Excellence, business value

11

Power AutomateFlows, approvals, AI Builder, process advisor

10

MLOps with GitHub ActionsModel deployment, testing & job automation

6

Microsoft TeamsMS-700 deployment, governance & administration

4

Power Pages, Chatbots & AzurePower Pages, conversational AI, cloud foundations

Recommendations

What leaders at Microsoft and Dell say.

A sample of 30 recommendations on LinkedIn from former managers and colleagues

Jason is proficient in the entire technical stack at Microsoft - Azure, M365, ADO, SharePoint, Power Platform, Power Automate, Power BI/Fabric, and AI/Copilot. He always gives 110% and is someone I trusted and relied upon to get the job done. I would hire him back immediately.

ElizabethDirector, Solution Strategy - Microsoft

Jason very quickly took reams of data and technical information and put it in formats that helped leaders make critical, time-sensitive decisions. He democratized the program data - accelerating progress by empowering others.

JoelDirector - Dell Technologies · PMP, Six Sigma Black Belt

One of the most skilled and knowledgeable Power BI resources I have ever worked with. He quickly understood our business needs and translated them into actionable insights through powerful visualizations and dashboards.

KesavakumarSr. Release & Software Quality - Microsoft

One of those rare individuals with strong people management skills and solid technical know-how. While an innovator, Jason is solution-focused, bringing a can-do attitude to all his work - truly a pleasure to be part of his team.

AnnSr. Program & Project Manager - Microsoft

His work always exceeded my high expectations, and he was always first to encourage a higher standard on the team and enforce best practices amongst his peers to show the way forward. Jason was also critical in helping keep the team motivated, well trained, and focused. As a team member, Jason also is just a great guy to work with, a real strong-culture guy. Laid back, easy to work with, highly adaptable to changing requirements and conditions, and just a guy you'd want to spend all day in the trenches with.

EvanSr. Program & Project Manager - Microsoft

Jason and I worked together in building the top internal portal for learning and readiness @ Microsoft. Jason managed our most technical web and content publishing customers. Jason was very adept at using the platform and focused customization to build the expected web experiences for very demanding customers. I would recommend Jason for challenging web production work and I would hire him if I had the chance.

GaryDirector, World Wide Learning - Microsoft

Let's Talk

Need an AI strategy that actually ships?

My current contract ends in mid-December, and I'm open to new roles, consulting engagements and full-time opportunities. If you want Copilot agents, automation, or intelligence built into your business - let's chat.

Jason@Pixel-Craft.com 425-516-8991 LinkedIn