Tambria Kemp — Product Manager

About the transition

Engineer → product manager

Six years at The Home Depot building products other people decided to build.

Product + engineering highlights

$72.8MAnnualized revenue impactOwned Mobile WebViews — 90% of homedepot.com traffic in the native app.
3 moFailed build to launchShipped Menovia on web, iOS, and Android after a year of prior failure.
90%+Test coverage at scaleCI/CD gates and error signals protecting transaction-critical journeys.
4Builds shipped end to endWebViews, Menovia, RiceBoard AI, Cre8 Visions automation engine.

Somewhere between writing the code and presenting results to executives, my role changed. I became the person who translated business goals into technical plans, defined the KPIs, ran the ceremonies, and decided what shipped next. I was already doing product management — now I want the title to match the work.

Engineers make exceptional product managers for a simple reason: we have lived downstream of every product decision ever made. We know what a vague requirement costs in rework, which features look small and take a quarter, and which look big and take a sprint. When I write a user story, engineering trusts the estimate behind it because I have given those estimates myself. That trust is not a soft skill — it is velocity.

AI is not a feature I bolt on, it is how I work. I use Claude, ChatGPT, and Copilot daily to draft requirements, prototype flows, and analyze feedback, and I shipped a subscription product with an integrated conversational AI companion for a client whose previous team spent a year and over $300K without delivering a working app. I delivered in three months. The PMs who thrive next will treat prompts, guardrails, and evaluation metrics as product requirements, not engineering details.

The kind of PM I intend to be is a partner, not a silo: a knowledgeable ear for the other product managers on the team, and an advocate for the engineers, because I have been the one on the other side of the backlog.

6 yrs enterprise engineering
$72.8M annualized impact
AI-native delivery

My work in video

Walkthroughs and product demos

Video 01

RiceBoard AI walkthrough

Placeholder — product demo coming soon.

Video 02

Menovia build story

Placeholder — product demo coming soon.

Video 03

How I run discovery

Placeholder — product demo coming soon.

The board

Click any card to open it

Now

Shipping this quarter

Validated, not started

Shipped & sunset

Receipts and lessons

Specs

Three decisions, written up the way I'd write them for a team

$72.8M

Annualized impact

$20M

Gross demand

90%

Of traffic on the platform

Context: Mobile WebViews is the rendering framework powering Home Depot's mobile commerce across iOS and Android. Every change ships into a transaction-critical path at enterprise scale.
What I owned: Product delivery for the platform, and the migration of revenue-critical journeys — product pages, search, checkout — into mobile WebViews. I was the translation layer between engineering architecture and executive roadmap decisions.
Tradeoff I'd defend: Shared platform over per-team speed. Onboarding every product team onto one framework was slower up front than letting each build its own. It's the reason a single extensibility improvement was worth $20M instead of one team's slice of it.
How I decided: SQL-backed dashboards in Embrace tracking load times, error rates, and crash rates. The KPI dashboard was the prioritization input, not a status report.
Outcome: $72.8M annualized revenue impact, $20M gross demand, 90%+ test coverage held through CI/CD quality gates.
What I'd do differently: Push the Webpack to Vite migration earlier. We carried slow builds for longer than the payback period justified.

v2 → v3

Full rebuild

Approver

My new role in delivery

Services automated first

Context: The first version of the agency worked. Demand was real, revenue grew, and delivery was entirely manual — which meant the ceiling was my own calendar.
The call: Shut it down instead of scaling it. Rebuild with automation as the production layer: AI Agents generate, I review and approve.
Tradeoff I'd defend: Narrower service menu, higher margin. I cut everything that couldn't be automated — email list building, email marketing, social — rather than keep revenue I'd pay for in hours.
Outcome: Delivery that scales past one person, with me reviewing output instead of producing it.
What I'd do differently: Automate before burnout, not after. The signal was in the timesheet a year before it was in my body.

3 mo

Discovery to launch

12+ mo

Prior team, no working app

2 → 3

Platforms covered

Context: A subscription product with a conversational AI companion at its center. After more than a year with a previous team, the client had an app that didn't work properly, a design she disliked, and a feature list that never got finished. It existed on iOS and Android only — no web.
What was actually wrong: Not just execution. Nobody had held the client's requirements as a spec, so a year of building drifted away from what she asked for. I started by writing down what she wanted and getting her to confirm it.
The call: Rebuild rather than patch, on one codebase shipping to web, iOS, and Android. Adding web was a scope increase that made the timeline shorter, because it removed the second and third parallel build.
Tradeoff I'd defend: Shared codebase over native polish. Three native builds would feel marginally better and would not have shipped. For a subscription product with no live users, shipping is the only thing that generates learning.
The AI work: I owned the experience end to end — prompt design, conversational flows, and guardrails — and treated all three as product requirements with acceptance criteria, not implementation details left to engineering.
Outcome: Live in three months across all three platforms, with the features the client originally asked for. Subscriber base has grown consistently since launch, with iteration driven by observed user behavior rather than feature requests.
What I'd do differently: Instrument the conversation quality earlier. I had usage data before I had a clean read on where the AI was actually failing people.

Changelog

Where shipped cards land

Download changelog

v4.0.0

Next release

Product manager

Breaking

Removes the engineer title. The responsibilities it described — roadmap, prioritization, release decisions — stay.

Feature

Merged the engineering, founder, and AI branches into main.

v3.0.0

2025 — now

Independent technical consultant — Cre8 Visions

Feature

Shipped Menovia in three months across web, iOS, and Android. Owned the AI experience end to end.

Fix

Rebuilt agency delivery on automation after v2 burned out the operator.

v2.0.0

2019 — 2026

Technical product manager / software engineer — The Home Depot

Impact

$72.8M annualized revenue impact. $20M gross demand. 13 internal awards.

Feature

Owned the WebViews platform carrying 90% of homedepot.com mobile traffic.

v1.0.0

2016 — 2019

Test lead — Twilio / Ionic Security

Feature

Led quality strategy across three concurrent enterprise security product teams, from test architecture through release readiness.

Impact

Defined the KPIs release decisions were made on — coverage, defect trends, regression stability.

v0.5.0

2015 — 2016

Engineer in test — Kabbage (acquired by American Express)

Feature

Validated web and mobile lending products for a high-growth fintech shipping on fast cycles under regulatory constraint.

Initial

Built the automation and CI pipeline behind it, and reviewed requirements before they reached implementation.

How I work

Read before installing

how-i-work.yaml

# Operating defaults. Adjust any of these to fit the team. discovery:  starts_with: "the problem in their words"  # Requirements are written down and confirmed with the stakeholder before  # development starts, so the build has something to be measured against. prioritization:  scored: "in the open"  # Inputs and assumptions stay visible, so a team can challenge the  # reasoning behind a ranking instead of only the ranking itself. metrics:  defined: "before the build"  # Load times, error rates, and crash rates are tracked in a dashboard and  # reviewed on a regular cadence. That data informs what gets built next. risk:  surface_at: "before budget moves"  # Technical risks are identified early and validated with a proof of  # concept whenever the answer would change the plan. scope:  cut_by: "what we can measure"  # Every release answers how success will be evaluated. Work that has no  # clear measure waits until it does. release:  gated_on: "readiness"  # Effort is sized with engineering and QA, coverage is held at the quality  # gate, and the decision to ship is made on evidence rather than the date. engineers:  default_stance: "advocate"  # I have been on the other side of the backlog. Teams do their best work  # when someone protects their focus and translates their constraints.

Stack

Filter by hat

Roadmapping RICE prioritization Discovery & validation KPI definition Experimentation Backlog management Agile / Scrum Stakeholder alignment User research UAT Jira & Confluence Figma React JavaScript SQL APIs CI/CD GitHub Selenium / Cypress Unit & API testing Webpack → Vite Embrace dashboards LLMs Prompt design Conversational AI Guardrails AI agent workflows AI performance metrics Claude Copilot Make.com

Reviews

Recognition, in other people's words

Awards and recognitions during six years at The Home Depot.

2019 — 2026

She identified the issue and joined the call within nine minutes. Tambria is what glued this together.

Nadine, colleague

I know I am not her only client, but she made me feel like I was.

Jennifer, benefits analyst · client

An excellent example of resolving a client's issue in a timely, effective and accurate manner. I love it.

Sal, vice president

She summarized every conversation, explained the next steps, and answered all my questions with patience.

Carol, client

She dropped everything to fix a large enrollment problem, then corrected our weekly file too.

Jennifer, benefits analyst · client

The full file

Every award and recognition email, unedited, in one PDF.

Download · PDF ↓

Install

Drag the app onto your team

tambria-kemp-4.0.dmg

Your team

Tambria.app

drag me over →

✓

Installed.

Available now. Reads the codebase, sizes the bet, tells you what to cut.

Email LinkedIn Download résumé