Matt Speck

Matt Speck

Matt Speck, photographed outdoors at sunset.
Seattle, 2026
AI Engineer Agentic Systems Seattle, WA

Big machine learning guy.

I enjoy building things at work and in my personal time. The bulk of my experience is in agentic systems—I've been building them since even before LLMs became big. I want to automate away all the boring tasks so that we can focus on the most engaging and creative parts of problem-solving. In my free time I love reading, watching films, playing pool, and going into the mountains.

Now

AI Engineer at Clear Fracture

AI-native systems for trusted data operations. Seattle, since August 2026.

Experience

2016 — present
2026 Present

AI Engineer

Clear Fracture Seattle, WA

Building AI-native systems for trusted data operations.

2020 5 yr 9 mo

Data Scientist

Cynnovative Washington, D.C.

Principal investigator and lead ML engineer on government-sponsored research programs, developing autonomous AI agents and machine learning systems for cyber operations and defense applications.

  • Secured a $1M contract win by spearheading technical proposal writing and leading end-to-end proposal strategy, beating out 30+ competitors.
  • Delivered an autonomous ML agent and a communication specification for DARPA multi-agent wargaming simulations featuring six competing agents, using PyTorch and reinforcement learning.
  • Advanced several multi-year ML research programs for cyber operations as PI, driving technical vision across transformer architectures, reinforcement learning, and computer vision.
  • Deployed containerized applications, MLOps pipelines, and APIs for government clients using Docker, FastAPI, and MLflow.
  • Shaped company procedures, business development strategy, and internal tooling as part of cross-functional startup leadership.
2018 1 yr 9 mo

Senior Consultant | Data Scientist

Booz Allen Hamilton Washington, D.C.

Built and deployed AI assistant products and data infrastructure for enterprise and federal clients.

  • Delivered an enterprise AI assistant for employee policy questions to Booz Allen staff, then adapted it for federal clients through domain-specific fine-tuning in Python and Docker-based pipelines.
  • Instrumented chatbot interactions and user queries at scale with Elasticsearch, enabling model training and evaluation against real usage.
  • Maintained Python APIs and data pipelines in containerized environments, improving platform reliability across client programs.
2017 1 yr 5 mo

Data Science Instructor

General Assembly Washington, D.C.
  • Independent Contractor2018
  • Lead Instructor2018
  • Assistant Instructor2017
  • Taught the full twelve-week immersive curriculum — Python, statistics, machine learning, capstones — to 30+ career-changing adult learners across three cohorts.
  • Designed and delivered a part-time course for 40+ government consultants, tailored to applied analytics in a federal context.
  • Ran a hackathon with Deloitte and a geospatial client project with TransitScreen; managed an assistant instructor, a PostgreSQL database, and the class GitHub organization.
  • Wrote course materials adopted by instructors globally for enterprise clients.
2016 3 mo

Bolivia Intern

Engineers Without Borders USA La Paz, Bolivia

Traveled to remote communities to assess and monitor irrigation, drinking water, and ecological latrine projects — four prospective partners assessed, three ongoing programs evaluated, legal documents translated from Spanish.

Education & research

2012 — 2016
2012 – 2016

B.S. Engineering Physics, minor in Economics

Fordham University — Bronx, NY

  • Undergraduate research assistant on whispering gallery mode biosensors: designed lab equipment in SolidWorks, ran acquisition and analysis in MATLAB and LabVIEW.
  • Best Undergraduate Student Poster, Connecticut Microelectronics and Optoelectronics Consortium.
  • Sigma Xi, The Scientific Research Honor Society · Dean’s List.
  • Engineers Without Borders, Fordham chapter — designed fish farms for construction in Uganda.

Publication Spatial multiplexing of whispering gallery mode sensors for trace species detection