Engineering executive · Founder · Seattle

I build products from zero—and the teams that scale them.

I co-founded a startup, helped transform engineering at Lowe’s, and now lead an 80-engineer organization at CVS Health. I work where product judgment, applied AI, and organizational change meet—and I still design and ship software myself.

Explore three stories View résumé

80 engineers
~3× shipping throughput
Monthly → 2-day deployment cadence
Knock → 2M+ units
2 granted U.S. patents

The short version

Three stories. One way of working.

Find the real constraint. Make the decision. Stay accountable for what happens next.

01 / Enterprise transformation

The bottleneck wasn't effort. It was how work shipped.

At CVS Health, my mandate spans an ~80-person engineering organization and a live healthcare estate. More code was not the answer; more dependable delivery was.

See the evidence

Constraint
Fragmented ownership, aging .NET systems, and competing release paths.
Decision
Rebuild accountability, modernize the platform, and standardize deployment.
Evidence
91 in-scope repositories moved to one deploy hub; shipped work grew ~3× quarter over quarter.

~3×Shipped / merged workQuarter-over-quarter engineering throughput, not a revenue claim.

02 / Zero-to-one founder

We changed the product until we found the real problem.

I co-founded Knock as CTO in 2014. Our first product found customers, but not a durable market. We pivoted toward the daily, revenue-critical work of apartment leasing.

See the evidence

Constraint
An initial product with customers but no durable market.
Decision
Pivot to leasing workflows and build product, infrastructure, team, and support from zero.
Evidence
A platform foundation that continued growing after I left in 2018.

2M+Apartment units servedKnock's later company scale, after my 2018 departure.

03 / Still a builder

Leading engineers shouldn't mean losing the craft.

I still take products from the first design decision through native clients, APIs, privacy, packaging, deployment, and maintenance.

See the evidence

Constraint
Useful products must work outside a demo, under real-world constraints.
Decision
Build local-first software, encrypted sharing, and session intelligence end to end.
Evidence
PaperPilot, Transfer, Playground, and VibeLogger are live product surfaces.

4Product surfacesIndependent PaperPilot family across native and web.

Evidence archive

The detailed evidence behind the work.

Seven case studies with direct responsibilities, engineering outcomes, and external corroboration where available. Company outcomes after my tenure are identified separately.

Operating at scale

Rebuilding a healthcare engineering organization while the business keeps running.

I reset ownership and leadership for an 80-engineer healthcare organization while the business continued operating.

80 engineers

CVS Health · Executive Director · Retail Health and Oak Street Health

Details & evidence

At CVS Health, I lead an ~80-person engineering organization—through a management layer, across US and Galway sites—spanning Retail Health and Oak Street Health. The mandate is ambiguous by design: strengthen internal ownership away from contractor-dependent delivery, modernize a large legacy .NET estate toward cloud-native and AI-native development, and leave behind a delivery model that keeps evolving after the transformation—all while the estate carries hundreds of millions of production calls each quarter.

Rebuilding the digital delivery system80 engineers across organization, governance, and platform modernization

First phase: Established a new leadership bench, hired 12 FTEs, moved critical ownership away from contractor-dependent delivery, replaced fragmented intake and gate-heavy governance with clearer product and engineering accountability, and began re-platforming toward cloud-native and AI-native development.

Scale of change: 30+ contractor transitions alongside continued business delivery, led through senior-manager, manager, and principal-architect layers across two sites.

Platform & delivery throughput

Make the platform the way the org ships.

I unified fragmented deployment paths so the organization could ship more reviewed work without simply writing more code.

~3× shipping throughput

CVS Health · Platform modernization · CAP + centralized CI/CD

Details & evidence

I moved an aging on-prem .NET estate onto a cloud-native platform (CAP) and standardized how the whole organization builds and ships. The headline is not a single re-platform—it is that shipping stopped being the bottleneck. Reviewed, merged, released work grew roughly 3× quarter-over-quarter while the volume of code authored rose only modestly. We were never short on effort; we were short on throughput, and that is the constraint I attacked.

Throughput, not just effortShipped work up ~3× quarter-over-quarter (peer-review throughput ~3.6×) while code authored rose ~1.3×

One deploy path: migrated 91 of 91 in-scope repositories off Jenkins, Octopus, and TeamCity onto a single standardized GitHub Actions deploy hub.

Retiring legacy risk: ~16 cloud-native services in production; legacy repository count cut 86 → 57; ~3.6 TB storage recovered and 91% of legacy cache traffic re-platformed.

AI-native engineering

Put AI on the critical path—under review.

I embedded AI into a human-reviewed delivery system, from reliable knowledge retrieval to repeatable remediation.

11 AI capabilities

CVS Health · AI-native SDLC + patient-facing delivery

Details & evidence

I made AI the connective tissue of the software lifecycle—not just code generation, but retrieving domain truth, drafting documentation, triaging incidents, and executing fleet-wide change under human review. The operating method mattered more than any model: discover, standardize, audit without changing anything, remediate in slices, then re-audit from re-runnable reports. Human gate on decisions, AI on execution, adversarial review always.

Engineering leverage: 11 internal AI tooling capabilities; build times cut from ~25 minutes to ~3; a hybrid-search knowledge base over ~1,900 internal documents so engineers and agents answer from documented truth.

Shipped to patients: virtual care live in 23 states; a scheduling-modernization program lifting digital booking conversion; and two of the estate's oldest, highest-risk back-office apps rebuilt engineering-led—feature-parity, 341 endpoints, zero breaking changes.

Reliability as an outcome: turned a live incident into durable self-serve observability—templatized dashboards, alerts, on-call, health endpoints, and OpenTelemetry across the estate.

Transformation leadership

Earn trust. Then change the system.

I earned trust by shipping the immediate need, then changed the architecture and release cadence of a distribution-center backbone.

Monthly → 2-day releases

Lowe’s Supply Chain · 50-engineer organization

Details & evidence

Founder-built platform

I learned company-building from both sides of the pager.

I co-founded the company, led the first product and engineering team, and helped pivot to the leasing problem customers faced every day.

Visit Knock

Co-founded Knock

Knock, Inc. · Co-founder & CTO · 2014–2018

Details & evidence

I co-founded Knock as CTO in 2014. Our first product found customers but not a durable market, so we pivoted to the daily, revenue-critical work of multifamily leasing. I led product and engineering from zero—designing workflows, writing the application, building the data and cloud foundation, hiring the team, onboarding customers, and taking support calls. That foundation kept compounding after my 2018 departure: Knock grew to serve 2M+ apartment units, raised $47M, and was acquired by RealPage in 2022.

My direct scope: Product strategy and design, application engineering, databases, infrastructure, hiring, onboarding, and customer support.

The product: A unified leasing front office spanning prospect communication, tour scheduling, workflow automation, integrations, and performance analytics.

Later company outcome: 2M+ units, 182 employees, and $47M raised by the 2022 RealPage acquisition; transaction terms were not disclosed.

External proof → GeekWire acquisition coverage

End-to-end product building

I still build the whole product.

I still design, code, package, and ship privacy-first native and web software rather than outsourcing the craft.

Explore PaperPilot

Native + web + cloud

PaperPilot · Independent product family · Native + web + cloud

Details & evidence

PaperPilot is where I keep the founder muscle active. I designed and built a privacy-first product family across native and web: PaperPilot organizes personal documents with on-device OCR and hybrid search; Transfer sends end-to-end encrypted files with passwords, expiration, and view limits; its Playground previews and prepares files in the browser; and VibeLogger records macOS work sessions, transcribes on-device, and turns them into searchable knowledge. The work runs from product design and clients through APIs, crypto, storage, AI integration, packaging, deployment, and operations.

Built across: SwiftUI iOS; native macOS products, including VibeLogger Interviews and Vibe Voice Input; Next.js web apps; Cloudflare Workers, D1, and R2; FastAPI OCR; auth and MCP; signed installers; and a multi-service control plane.

Innovation Labs — AI, 3D & home data

From weekend prototype to production platforms.

I turned prototypes into production platforms spanning AI assistants, 3D shopping, and home data.

See the official projects

5,500+ product pages

Lowe’s Innovation Labs · Applied AI · 3D commerce · Home Data · 2019–2025

Details & evidence

Across Lowe’s Innovation Labs I ran charters spanning applied AI, 3D and spatial commerce, and a home-data platform—each starting from a working artifact and ending in production. I designed and built Lowe’s Product Expert over a weekend, then carried it through brand, legal, and the OpenAI partnership to become the first custom GPT for home-improvement retail—which helped lead to my Core AI mandate.

Artifact-first AIBuilt Lowe’s Product Expert over a weekend and drove it through brand, legal, and the OpenAI partnership—the first custom GPT for home-improvement retail

Platform compounding: led the 3D commerce foundation—asset pipeline, delivery layer, and embeddable viewer—live on 5,500+ product pages by November 2021 and later a base for Open Builder and spatial experiences.

Granted IP: Named co-inventor with six colleagues on two Lowe’s-assigned U.S. patents for modeling physical products as virtual objects and arranging them in 3D environments: US 12,189,915 and US 12,211,161.

Data platform to customer product: owned Lowe’s Home Data—a $10M+ data and AI platform with a pipeline spanning 140M+ U.S. homes and agentic itemization workflows—led through Naveen Jammula, a manager reporting directly to me. It launched November 2024 as the Lowe’s Digital Home Platform; Lowe’s showcased its next expansion at CES 2025.

External proof → Lowe’s Product Expert

External proof → Lowe’s launch announcement

External proof → Lowe’s at CES 2025

Team corroboration → Naveen Jammula

The through-line

Founder to Fortune 50, without losing the craft.

My best work sits where product instinct, engineering depth, and organizational judgment meet. At Knock, that meant doing every job needed to find product-market fit. At Lowe’s, it meant leading Innovation Labs, Supply Chain, and Core AI. At CVS Health, it means rebuilding platforms and practices around cloud-native and AI-native development—at healthcare scale.

2014Co-founded KnockZero-to-one CTO

2018Joined Lowe’sThree charters, three promotions

2025Chose healthcareMission at enterprise scale

How I operate

Leadership is a way of building.

Make the future concrete.

A working artifact changes a strategy conversation. I use code to collapse ambiguity, then do the organizational work required to make the idea real.

People over pedigree.

I hire for raw talent, growth mindset, and ownership. The best organizations are designed so strong people can see the whole problem and act on it.

Match rigor to the moment.

The craft is knowing when to move fast and when to do it right. Prototype speed and production discipline are different modes, not opposing beliefs.

Reduce organizational friction.

Architecture and organization shape each other. I design systems, ownership boundaries, and decision paths that let teams move without waiting for a gatekeeper.

The person behind the work

Meaningful work is a design constraint.

After a long stretch in hospital waiting rooms with someone I love, I chose healthcare for the next chapter of my career. It made the system personal—and made its opportunities impossible to see as abstractions.

I want difficult problems, capable and kind people, and products whose usefulness is easy to explain. Excellent systems matter because of what they let people do.

Writing

Ideas that need more room.

Essays on agent infrastructure, AI-native engineering, and the organizations that build them.

Agent memory · Field reportWhat 1000+ coding agent sessions taught me about LLM memory.

Context is not continuity, transcripts are not memory, and durable state should survive the agent.

Agent memory · GuideWhat is agent memory?

Agent memory is durable state that helps an AI agent preserve useful facts, decisions, preferences, and project knowledge across sessions.

Agent memory · ConceptsAgent memory vs. context: why bigger context windows do not create continuity

Context determines what an AI model can see now. Agent memory determines what should remain useful across sessions.

Agent memory · ConceptsAgent memory vs. transcripts: session history is evidence, not memory

Agent transcripts preserve what happened. Useful memory selects and maintains what a future agent should know.

Agents & architectureThe semantic layer agents are missing

Software built for agents needs to expose meaning, intent, and safe action—not merely endpoints.

Agent memoryMemory is infrastructure, not a prompt trick

Useful memory is a lifecycle: capture, retrieval, context budgeting, correction, and recovery.

Engineering organizationsWhat AI-native engineering actually changes

Coding assistance is the visible edge. The deeper change is how organizations express and move intent.

Visit writing

Nights and weekends

I still ship.

Small products keep my technical judgment honest. They force ideas back through real constraints: users, architecture, maintenance, and failure modes.

Open source

pi-memory

A plain-text memory layer for the Pi coding agent, with optional semantic and hybrid retrieval. As of September 7, 2026, it ranks #18 of 5,402 packages in Pi’s download-sorted catalog and is the catalog’s highest-downloaded memory package, at 37.1K monthly downloads (151 stars, 31 forks).

Open project Open source

agentmemory

Portable, inspectable memory for coding agents including Claude Code, Codex, Cursor, and Agent. Published as myagentmemory on npm; as of September 5, 2026: ~1.2K monthly downloads, 19 stars, and 3 forks.

Open project Product family

PaperPilot

A privacy-first product family spanning personal documents, encrypted file transfer, browser file tools, and local-first work-session intelligence.

Open project

Earlier open-source workContributor to pgcli, the PostgreSQL CLI, acknowledged in the project’s AUTHORS file.

A good conversation

Interesting problems deserve builders who can lead.

If you're building an engineering organization, modernizing a platform, or turning applied AI into a real product, I'd enjoy comparing notes. The best conversations start with the problem, not a job title.

Start a conversation