Matthew Keas, Developer in Houston, TX, United States
Matthew is available for hire
Hire Matthew

Matthew Keas

Verified Expert  in Engineering

Distributed Systems Developer

Houston, TX, United States
Toptal Member Since
April 28, 2015

Matthew is a machine learning and generative ML expert working as a full-time engineer at Google (Cloud org). He is an accredited ML leader at Google. Matthew has a background in distributed and highly available systems, app development, and multi-cloud architectures.


Machine Learning, Generative AI, Graph Neural Networks...
GFT Group
Docker, Oracle RDBMS, Java, Google Cloud Platform (GCP), Nomad, Kubernetes, Go
StackWave, LLC
Windows, Linux, Java, Octopus Deploy, Oracle RDBMS, Event Sourcing, TypeScript...




Preferred Environment

Visual Studio Code (VS Code), Vi, Git, Linux, Go, MacOS, Google Cloud Platform (GCP), Python, TypeScript, C++

The most amazing...

...thing is being trusted to work on the bleeding of ML systems and applied AI engineering at Google.

Work Experience

AI Engineer

2022 - PRESENT
  • Served as an accredited ML/AI leader in Google's applied AI teams.
  • Produced tools and internal documentation used by over 5,000 Googlers as of April 2024.
  • Delivered the most bleeding-edge ML implementations for HCLS, retail, telecommunications, and manufacturing customers.
Technologies: Machine Learning, Generative AI, Graph Neural Networks, Neural Machine Translation, Neural Networks, Deep Neural Networks, Knowledge Graphs, Scalable Architecture

Senior Architect

2019 - 2022
GFT Group
  • Migrated enterprise monolith software stacks to cloud-native distributed models for clients, including JPMorgan.
  • Wrangled code and clusters for greenfield enterprise software for both internal an external-facing needs.
  • Configured and augmented service mesh integrations with homogeneous and hybrid-cloud applications.
Technologies: Docker, Oracle RDBMS, Java, Google Cloud Platform (GCP), Nomad, Kubernetes, Go

Senior Software Engineer

2017 - 2019
StackWave, LLC
  • Analyzed genomic information and coded out-of-the-box solutions for laboratory information management systems for use in biotech firms and drug discovery.
  • Created abstractions to database layers, providing mux/demuxed protocols for GraphQL and RESTful endpoints.
  • Implemented new TypeScript-based build systems and front-end compilation to catch more programmatic and logic errors.
  • Led projects/teams and interfaced with company clients to lead efforts on custom development for their LIMS-deployments.
Technologies: Windows, Linux, Java, Octopus Deploy, Oracle RDBMS, Event Sourcing, TypeScript, JavaScript, PostgreSQL, C#, .NET

Chief Instructor

2014 - 2017
The Iron Yard
  • Mentored beginner coders through a twelve-week programming bootcamp.
  • Mentored other instructors in the art of pedagogy and leading a classroom.
  • Built internal tools and software, spoke at conferences, and wrote books.
  • Hacked on side projects, such as Verlet Physics Engines, using Canvas, Web Audio API, and Web Workers, as well as developing projects and curricula for the bootcamp.
  • Organized community events and ran free community coding workshops.
Technologies: Continuous Delivery (CD), Continuous Integration (CI), Kubernetes, Docker, SCSS, Firebase, Parse, DigitalOcean, Heroku, Angular, Backbone.js, React, Node.js, JavaScript

Senior Software Engineer

2012 - 2014
Pariveda Solutions
  • Led development teams as a JavaScript and front-end specialist.
  • Contributed to "best practices" and led workshops internally on JavaScript.
  • Wrote technical articles and blog posts and started authoring books.
  • Spoke at technical conferences and meetups, representing Pariveda.
  • Built an enterprise metal-building configuration app in the browser (ASP .NET MVC 4.5, Backbone, Raphael (SVG)) for NCI Group.
  • Researched and documented financial and title-insurance data discrepancies for Stewart Title.
  • Built Windows 8 apps for Microsoft and Halliburton.
  • Developed a custom, responsive front-end single page app platform for large medical research institution Baylor College of Medicine.
  • Pioneered research into mobile enterprise app platforms (MEAPs) for clients such as Kony Studio (certified), Titanium Appcelerator, and Xamarin.
  • Updated large-scale news-delivery services (updated custom PHP platform) for Hearst; contributed to documentation of the platform during the discovery phase, documenting findings and lessons learned.
Technologies: C#, Xamarin, Appcelerator Titanium, Kony, Amazon Web Services (AWS), Azure, Node.js, JavaScript, Python, .NET

Lead Developer

2011 - 2012
  • Revamped the existing custom PHP platform into CodeIgniter.
  • Built custom calendaring systems and UI for scheduling second-opinion appointments online, including scheduling "recurring events" with start and end dates.
  • Standardized the REST API and helped scale services on Rackspace; implemented load balancers with Nginx and Redis.
  • Pioneered the use of Node for live text, video, and audio chat using TokBox and Twilio; also implemented file-sharing.
  • Developed a live-monitoring dashboard for use by the team that reported live sessions, and alerted the team if a doctor or client's machine wasn't working. This made it easy to know someone's computer didn't support the required technologies before they were aware there was an issue. We would be able to initiate calls to customers, providing a great customer service experience.
  • Researched and developed iPad and iPhone apps with Titanium Appcelerator, published to the App Store; allowing 2nd.MD to stay relevant in the mobile space.
Technologies: Twilio, Redis, Solr, Rackspace, jQuery, Appcelerator Titanium, TokBox, WebSockets, Node.js, MySQL, JavaScript, CodeIgniter, PHP

On-demand pharmacy delivery and order management system.

Tech: React Static Boilerplate, React, Redux, Redux-Form, Redux-Saga, Webpack, PostCSS, Mocha/Chai for TDD. AWS Managed Kubernetes for Node.js API Docker deploys. AWS S3 for static asset hosting. Stripe for payment processing.

Magnolia Market/Fixer Upper Mobile App
Swift app for Magnolia Farms' eCommerce shopping platform, as made famous by HGTV's Fixer Upper.

Tech: Firebase SDK, TypeScript, React Native, Mobile analytics, Swift, Shopify integration, Apple Pay, integrated custom user properties/events/funnels for analytics.

Planned "large surface area" heuristics to look for with Google BigQuery (exporting data from Firebase) to understand widespread user habits and product/category performance.

Arbiter – A Live Coding Browser IDE
Live coding environment with offline + ECMAScript 2016/2017 + Flow (type annotation) support. Uses an embedded Babel instance compiled for the browser environment to compile user's code on the fly.

• Offline IndexedDB and LocalStorage cache allows for the fastest, most-natural visual response of keypresses.
• Embedded proxying of JS errors and TDD calls to the UI for intuitive and unobtrusive debugging.
• Support for saving/embedding links in Google URL shortener (for small enough files).
• No database needed—code is embedded in the URL, easily shared without overwriting the presenter's code.
• Great tool for testing library and algorithm integrations—support for `require()`-ing npm packages via browserify-ing proxy.

Tech: HTML5 low-level execution and Frame message-passing via streaming/observables; Flow + Webpack + Babel + ES6/7;

StackWave, LLC

StackWave builds custom .NET platform deployments to enable drug research teams to collaborate and speed up analysis and analytical data mining of genomic and protein-folding data. Powered by .NET, a custom platform and UI components, and a heavily optimized data layer on top of PostgreSQL.

Many customers have custom deployments that I have worked on, including several IPO'd genomics and drug research companies and the largest non-profit research institutions in the US.

2nd.MD Node/Chat/Video and Mobile Apps
Tech: CodeIgniter, Apache Server, PostgreSQL, and Solr for full-text search. The real-time chat components were built on Node.js Streams and WebRTC, and the UI components were coded primarily with Backbone and jQuery. The mobile apps for iOS (iPhone and iPad were supported) were built using Appcelerator Titanium and the JS-bindings to the Native iOS platform.

NCI Building Systems

Tech: ASP.NET MVC, MS SQL, IIS dedicated host. Front-end: Backbone app with interactive SVG drawings driven by Raphael.js. Internationalization and single-page app framework designed for the project.

Offloaded processing of engineering workflows for the specification, validation, and calculations involved in defining, geographically placing, and fine-tuning of metal structures against building codes and physical loads/requirements.

Baylor College of Medicine

Discovery platform: An open marketplace for BCM network scientists to leverage compute resources, paid for with grant funds.

Tech: Node.js, PostgreSQL, custom back-end CRM and eCommerce integrations. Backbone.js, responsive front-end with animations.

Configured deployments with Docker, configured and shared images in Docker Hub.

Halliburton Workflows

Windows desktop workflow and task management app for Microsoft to pre-install on large hardware order for Halliburton, integrating with legacy workflow SOAP endpoints. Implemented live tiles, badges, toast notifications, background tasks, and lock screen updates.

Tech: Windows Desktop SDK, notifications using Azure Notification Hubs, legacy SAP web service integration via SOAP endpoints proxied through Azure cloud services, Docker configurations for Node.js services.

Are You Facing North?

Interactive experience integrating video streams from a non-linear documentary for submission to Sundance Interactive.

Tech: Heroku deployment pipeline, functional reactive programming, sheets.js, responsive WebGL UI with physics-engine.


Contracted by an NFL veteran of both the Atlanta Falcons and Cincinnati Bengals, I collaborated with a designer to launch the first version of Fanflow, a marketplace for autographed sports memorabilia straight from pro-footballers, enabling pros to sign and distribute goods directly to fans.

Tech: Preact/React isomorphic app, TypeScript, Node.js, Zeit Cloud, Docker build, GraphQL APIs hosted over Google Cloud SQL (PostgreSQL); serverless endpoints and webhook integration; Stripe and messaging (SMS/Email) integration; tax record and background check verification.

The project was put on pause in September 2017 to allow the owner to try other marketing approaches and business deals.

Clan is a super succinct, no-dependency set of utilities written in TypeScript with a slightly opinionated collection of features that integrate particularly well when used together.

Explores approaches in functional data structures, virtual DOM, multi-proc data streams, pipelined web server streams over low-level Node.js APIs

BatchQL is a language-level query optimizer for GraphQL. It parses GraphQL queries into an Abstract Syntax Tree, and magically binds logical queries and minimizes payload and chattiness to speed up parallel requests to a GraphQL endpoint.

Uses some pretty complex compiler logic to achieve this at an insanely fast speed using a technique called parser-combinators.

Faster than Apollo's query-batching implementation.

Flyboy Analytics

Hillsboro Aviation flight schools contracted my clients to build an in-flight device to automatically track and report analytics and scorings on flight students' sessions, as well as integrate some machine learning models and statistics across whole schools and instructors to rank and score general student and staff success rates.

Software included full A/V recording + playback, pitch/yaw/heading/speed/etc tracking with high precision 6-axis gyro and sensors, plus interactive highlights of trouble spots and study points backed by data and heuristics from student sessions.

Goal: to accelerate student learning success in flight school scenarios.
2010 - 2011

Professional (Certification) Degree in Information Systems Security (INFOSEC)

University of Texas - Austin, TX

2007 - 2011

Bachelor of Science Degree in Computer Science

University of Texas at Austin - Austin, TX


Kony Certified Developer

Kony App Platform


Information Systems Security (INFOSEC) Professional Certification

Committee on National Security Systems, National Information Assurance Education & Training Program Office within the Information Directorate at the NSA


Health Information Privacy and Security Professional

Office of the National Coordinator for Health Information Technology, Department of Health and Human Services


React, Socket.IO, Node.js, LocalStorage, X (formerly Twitter) API, Facebook API, SoundCloud API, Velocity.js, jQuery, Restify, HTML5 Canvas, Backbone.js, Backbone.Marionette, Vue, API Development, WebRTC


Wintersmith, Metalsmith, Appcelerator Titanium, GreenSock, Terraform, Microsoft Unity, Git, Mercurial, Sublime Text, TokBox, Solr, Docker Hub, Docker Compose, Kafka Streams, Mesos, Google Kubernetes Engine (GKE), Docker Swarm, Azure Kubernetes Service (AKS)


Express.js, Mithril.js, CodeIgniter, Hapi.js, ASP.NET MVC, React Native, Django, Next.js, Angular, .NET, Spring


C#, JavaScript, PHP, TypeScript, Java 8, Java 9, Java 6, Python, Java, Groovy, Go, SCSS, Ruby, Haskell, C++


Functional Reactive Programming, Promise, REST, Dependency Injection, Functional Programming, Continuous Integration (CI), Continuous Delivery (CD), Event Sourcing


DigitalOcean, Parse, Kony, Xamarin, Heroku, Firebase, Linux, Kubernetes, WordPress, Amazon Web Services (AWS), Azure, Rackspace, Twilio, Windows, Google Cloud Platform (GCP), Oracle, Docker, Apache Kafka, Visual Studio Code (VS Code), MacOS


MySQL, PostgreSQL, Google Cloud, PostgreSQL 10, Graph Databases, Databases, MongoDB, Redis, Oracle RDBMS, Docker Cloud


Physics Simulations, Verlet Engines, 2D Games, Mobile Web, WebSockets, Software Development, Front-end Development, Freelancing, Enterprise Systems, Computer Science, Compilers, Distributed Systems, Machine Learning, Vi, Octopus Deploy, Nomad, Apps, Spring Dependency Injection (DI), Enterprise Java Beans (EJB) 3, Kubernetes Operations (kOps), Generative AI, Graph Neural Networks, Neural Machine Translation, Neural Networks, Deep Neural Networks, Knowledge Graphs, Scalable Architecture

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.


Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring