Skip to main content
Eduardo Muth MartinezMS · PMP
Menu

Technical program and product leader

Build the system. Change the outcome.

At Expedia Group, I lead development and own supply and content-quality platforms across more than 20 teams. I also serve as an AI Champion. Outside Expedia, I operate three consumer products and build Formation’s expert network for agents and Mallanet’s free disaster-response infrastructure.

Read the stories Work with me

Chicago area · Working globally

A system trace links enterprise platforms, consumer products, agent workflows, and disaster-response infrastructure.

Expedia Group · Two areas

Trustworthy property content and practical AI enablement.

At Expedia, inaccurate property details can undermine a booking. Teams adopting AI need concrete examples, useful guidance, and a place to test both against real work.

01

Expedia Group · Content quality

Helping travelers trust the property information they see.

Property information moves from uncertain to verified.

Three states show a beachfront claim, parking fee, and current amenities moving from uncertainty through review to verification.

An inspection sheet resolves property claims into clearer traveler information.

  1. Context

    A changing claim

    Property details change. A property may not be as close to the beach as its listing says. Amenities disappear, parking fees change, and content falls out of date.

  2. The work

    Evidence, reviewed

    I lead the development and own platforms that proactively detect, flag, and correct content errors. That can mean checking whether a property is really beachfront or identifying the exact parking fee.

  3. Outcome

    A clearer decision

    Before booking, travelers can see more accurate location, parking-fee, and amenity information.

The platform work spans more than 20 teams and keeps verification focused on what the traveler needs to make a decision.

What changes

The platforms catch and correct outdated property details before a traveler books.

02

Expedia Group · AI Champion

Move AI from abstract interest to usable practice.

AI adoption starts with real work and becomes a repeatable practice that can evolve with the technology.

A three-step workshop board connects a decision map, a team pilot, and a repeatable cycle for documenting, sharing, and updating pilot evidence.

An Expedia adoption sheet moves from a real task to a team test and a repeatable practice that can evolve with the technology.

  1. Context

    Find the real work

    I start with the decisions, handoffs, and repeated tasks where a team can test whether AI is actually useful.

  2. The work

    Make it concrete

    As one of Expedia Group’s AI Champions, I organize hackathons and demonstrations and help teams test practical tools in their day-to-day work.

  3. Outcome

    Build a repeatable practice

    I document pilot evidence in best-practice guides, reusable examples, and tools that others can test, update, and improve.

Hackathons and demonstrations test AI against real work. What holds up is documented in guides, examples, and tools teams can revisit and update.

What changes

This gives Expedia a repeatable way to test AI on real work, preserve what teams learn, and keep its guidance current as the technology changes.

Products in operation

I also build and operate products end to end.

Across the portfolio, I handle discovery, UX, mobile, backend, analytics, monetization, and app-store operations.

03

From inventory to a decision

Clueless: Outfit Planner · Live on iOS and Android

Turn a full closet into a plan for one real day.

A full closet can still leave someone unable to decide what to wear, especially when they cannot see how the pieces work together.

Wardrobe cards are grouped into an outfit, then the outfit is assigned to a date in the planner.

Clueless combines owned pieces into an outfit and assigns it to a day in the planner.

  1. Context

    See the inventory

    The app records the pieces someone owns in one searchable wardrobe.

  2. The work

    Build outfit options

    Owned pieces combine into outfits that can be compared, saved, and used again.

  3. Outcome

    Plan a day

    The selected outfit is assigned to a date in the planner.

I build and operate Clueless across mobile, backend, subscriptions, analytics, monetization, and app-store releases.

What changes

Someone can see what they own, combine pieces into outfits, and plan what to wear on a specific day.

04

From intention to action

Ocho · Live on the App Store

A distant goal becomes this week’s three actions.

Long-term goals are easy to defer when they never reach the calendar. Ocho links each one to the current week.

A long path from a distant goal shortens into one week and ends in three clearly labeled actions.

A seven-layer Goal Stack narrows into three actions placed on this week’s calendar.

  1. Context

    Name the distant goal

    Purpose and a High Hard Goal anchor the top of a seven-layer Goal Stack.

  2. The work

    Choose this week’s outcome

    Annual, quarterly, and monthly layers connect that goal to one outcome for this week.

  3. Outcome

    Take three actions

    The weekly ritual selects three concrete actions and places them on the calendar.

I built and shipped Ocho as an iOS app around a seven-layer planning model.

What changes

The weekly Goal Stack contains three scheduled actions tied to the larger goal.

05

From recognition to recall

Banded IELTS · Live on the App Store

Practice the IELTS words you recognize but cannot yet recall.

Banded turns a pasted vocabulary list into an offline drill. The learner recalls the meaning before revealing the answer, and SM-2 schedules the next review from the difficulty rating.

Three states show a built-in ‘assess’ card with its answer hidden, the definition and example revealed, and the SM-2 choices that return it to the due queue.

One study card moves from recall to a difficulty rating, then returns to the queue with a new review date.

  1. Context

    Import the study list

    Paste a list or import a CSV or TSV file. Banded turns it into a study deck that remains available offline.

  2. The work

    Recall before revealing

    One of Banded’s built-in cards shows “assess” while keeping the definition and example hidden. The learner answers before revealing them.

  3. Outcome

    Schedule the next review

    After the reveal, Hard, Good, and Easy set the next interval. The card returns to the due queue with the answer hidden again.

I built and shipped Banded as an iOS app with list imports, offline study, and an SM-2 review flow.

What changes

The next session contains only the words due for review, even offline.

Independent systems

Knowledge for agents and disaster-response systems.

Formation prepares attributed expert knowledge for agents. Mallanet turns systems used in Venezuela and Colombia into free infrastructure for future disaster response.

06

Independent project · Formation

Give your agent someone worth following.

Formation connects authorized expert sources to the agents that follow them.

Three states show authorized sources, maintained agent-ready knowledge, and attributed retrieval through an agent connection.

Expert sources move through preparation and attribution into a controlled agent connection.

  1. Context

    Authorize the sources

    Selected experts keep publishing where they already do. Formation works from the sources they authorize.

  2. The work

    Prepare the knowledge

    Formation translates, checks, versions, and attributes the forms that agents can use while the expert keeps one body of work.

  3. Outcome

    Connect the agent

    An agent can discover relevant experts, explain the match, and retrieve permitted knowledge with its source and revision attached.

Formation is now in free public beta, with me as the first expert.

What changes

A person stays in control while their agent gets current knowledge from experts they trust.

07

Public-interest systems · Venezuela and Colombia

Make critical response information searchable when ordinary systems fail.

Scattered reports become a shared, reviewable response registry.

Three states show fragmented records moving through classification and coordinator review. The final state shows the Venezuela deployment connecting thousands with missing loved ones and aid and the Colombia deployment going online in about two hours and mobilizing thousands of volunteers.

Paper lists, message threads, and spreadsheets become one classified registry used in Venezuela and Colombia.

  1. Context

    Records across disconnected systems

    After an earthquake, families, hospitals, and volunteers need current information while normal systems are disrupted. Names, needs, shelter updates, and supplies arrive through paper lists, message threads, and separate spreadsheets.

  2. The work

    Classify and review

    I led the build of one registry to receive, classify, and track reports. It keeps each source attached and routes possible duplicates to coordinators for review.

  3. Outcome

    Deploy and mobilize

    We adapted the system for Colombia and put it online in about two hours. It became the response’s largest volunteer-coordination platform. It mobilized thousands through a global network and matched them to needs in the field.

Mallanet makes the registry available as free, reusable infrastructure for future disaster response.

What changes

The Venezuela platform connected thousands with missing loved ones and aid. It supported hospital patient-information workflows. In Colombia, thousands of volunteers mobilized through a global network.

Selected technologies

The tools behind each project.

The stacks change with the work. Clueless needs mobile subscriptions and image processing; Mallanet needs maps, queues, and production telemetry; Formation needs typed workflows and execution evidence.

01

Clueless

The mobile app and its production backend cover subscriptions, authentication, image processing, analytics, and AI-assisted wardrobe work.

  • Flutter
  • Dart
  • Riverpod
  • Python
  • FastAPI
  • Pydantic
  • Firebase
  • Firestore
  • BigQuery
  • Redis
  • OpenAI
  • Anthropic
  • RevenueCat
  • Sentry
  • Cloud Run

02

Formation

Formation uses typed files, schemas, and MCP to record workflow state, approvals, evidence, and verification.

  • TypeScript
  • Node.js
  • React
  • Vite
  • YAML
  • JSON Schema
  • AJV
  • MCP
  • Anthropic SDK

03

Mallanet

The disaster-response platform serves public maps, intake and admin tools, queues, and a separate metrics and log stack.

  • TypeScript
  • React
  • Next.js
  • Express
  • PostgreSQL
  • Drizzle
  • Valkey
  • BullMQ
  • Leaflet
  • Docker
  • Caddy
  • k3s
  • Prometheus
  • Grafana
  • Loki

04

Furbo

Furbo is a collaborative project with my brother. The product uses a React app, Supabase data, Cloudflare infrastructure, analytics, and motion.

  • React
  • TypeScript
  • Vite
  • Supabase
  • PostgreSQL
  • Cloudflare Workers
  • D1
  • PostHog
  • Zustand
  • Motion

05

Web and editorial systems

The portfolio and launch sites are static-first Astro builds deployed through Cloudflare, with browser, accessibility, link, and performance checks.

  • Astro
  • TypeScript
  • Tailwind CSS
  • Cloudflare Pages
  • Wrangler
  • Playwright
  • axe-core
  • Lighthouse CI

06

Release and quality tooling

I maintain Python and TypeScript MCP servers and CI workflows for store metadata, builds, screenshots, QA, and releases.

  • Python
  • TypeScript
  • MCP
  • GitHub Actions
  • Doppler
  • App Store Connect
  • Google Play
  • pytest
  • Vitest
  • Flutter integration tests

Career

From restaurant deployments to enterprise platform leadership.

Silverware, LocusView, and Expedia form the primary line. Clueless, Formation, and Mallanet run alongside it as products and independent systems I build and operate.

From Silverware field service to LocusView implementation, Expedia platform leadership, and independent systems.
  1. 01
    Silverware POS

    Field service, restaurant deployments, and senior project management.

  2. 02
    LocusView

    Solution design, integrations, and enterprise rollout.

  3. 03
    Expedia Group

    Program and platform leadership across more than 20 teams.

  4. 04
    Independent systems

    Clueless, Ocho, Banded, Formation, and Mallanet.

Primary experience

–

Silverware POS

Field Service → Implementation → Senior Project Management

Progressed from field service to senior project management, leading restaurant technology deployments from discovery and staging through rollout and adoption.

–

LocusView Solutions

Technical Project Manager

Led customer-facing enterprise implementations from solution design through integrations, workflow configuration, and rollout.

– present

Expedia Group

Program Manager II → Program Manager III

Lead the development and own supply and content-quality platforms across 20+ teams, including proactive error detection and correction. Separately, drive AI adoption within my organization as an AI Champion.

Parallel work

– present

Clueless Creations

Founder

Build and operate consumer subscription products across mobile, backend, analytics, monetization, and app-store operations.

– present

Formation

Builder

Built a typed agent-workflow engine with 97 reusable workflows across 15 domains, including approvals, evidence, and independent verification.

– present

Mallanet Inc.

Founder and Board Member

Founded and led free disaster-response infrastructure that connected thousands with missing loved ones and aid, supported hospital patient-information workflows, and, in Colombia, mobilized thousands of volunteers through a global network.

Writing

Notes on products, systems, and work that compounds.

The Compounding Founder is where I document decisions, experiments, and lessons from building.

01

The Money Decision You Make Before You Build

Choose how your app earns before you design a single screen, or you'll be rebuilding it later.

 02

Rich Sutton is right. That’s a problem.

If intelligence requires machines that never stop learning, what exactly are we aligning?

 03

How to Get Your First 1,000 Users

The finale. The 3 channels that actually work for a solo non-technical founder, and the ones that will waste your launch week. Plus: the day-by-day playbook for the week everything changes.
Read all writing

What I’m following

What’s shaping my thinking.

Books, conversations, people, and tools I return to, with a note on why each one is here.

01
Reading

Co-Intelligence

Ethan Mollick

The book I recommend first when someone asks how to work with AI without turning the conversation into a technical lecture.

 02
Reading

One Useful Thing

Ethan Mollick

Grounded experiments and observations from someone who uses the tools daily. The back catalog is unusually useful.

 03
Listening

How I AI

Claire Vo

Real people walking through real AI workflows. I care much more about the work on screen than another round of predictions.

 04
Listening

Skill Issue

Andrej Karpathy on No Priors

A useful conversation about code agents, parallel work, AutoResearch, and what changes when experimentation gets much cheaper.
Explore the library

About

I lead programs and still build.

I’m Eduardo Muth Martinez, MS, PMP. I started in restaurant field service, moved through enterprise implementation, and now lead platform work at Expedia. Building Clueless, Ocho, Banded, Formation, and Mallanet keeps me close to the product and engineering decisions in each system.

Portrait of Eduardo Muth Martinez.
Eduardo Muth Martinez · MS · PMP

Contact

Need technical leadership for a product, platform, or AI program?

Email me LinkedIn ↗ GitHub ↗