Joe Waugh
I design and ship production ML for operations, and the data platforms it runs on, from first idea to the business case that funds it.
I'm a senior machine learning engineer at ThredUp in Scottsdale, where I build computer vision and ML for our distribution centers on a Databricks lakehouse, from first idea to production. My garment-measurement work is featured in Databricks' ThredUp customer story. This year I led the Algorithms team for a quarter, wrote the team's AI roadmap for 2027, and started building the shared infrastructure that lets teams run AI agents safely.
Before that I spent three years at FedEx. I started in operations research, building the tooling behind a $1.2B capital plan, then moved into data science, where I built the fraud alerting platform that kept a $300M customer from leaving. I started my career at State Farm, running experiments on telematics products.
I finished my undergraduate degree in business data analytics at Arizona State in two years, at 19, and earned an MS in computer science from Georgia Tech while working full time.
Outside my day job, I mentor students through Georgia Tech and Arizona State, serve as a peer mentor on AI, and take on a little advising.
Selected results
At ThredUp
- $1.4M
- a year in labor saved by automating garment measurement.
- 98%
- of pant measurements automated, up from 80%.
- 49%
- of item intake fully automated, up from 42%.
I also build end-to-end, self-learning ML pipelines for critical operations workflows.
At FedEx
- $300M
- customer kept after my fraud alerting platform found where packages were disappearing. Read the story.
- $1.2B
- capital plan built on my planning tooling.
- $3M
- in projected revenue from an image model that catches missed surcharges.
- $2M
- a year saved by auditing proof-of-delivery photos. Read the story.
- 5%
- more accurate delivery-date windows on FedEx.com during the holiday peak.
What I work with
- ML systems
- Computer vision (pose estimation, measurement, flaw detection), forecasting, fraud and anomaly detection, generative image pipelines, local LLMs and agents
- Data platform
- Databricks lakehouse: Spark and PySpark, Delta Lake, Unity Catalog, Jobs, notebooks; SQL
- Serving and MLOps
- Kubernetes model serving, AWS SageMaker, latency tuning, monitoring in production
- Decision support
- Operations research, capital planning, business cases, roadmaps for executives
Career
- Work
- Study
- Mentoring and hiring
Drawn to scale. Select a bar to jump to that role. Scroll the chart sideways to see earlier years.
- ThredUp, Senior Machine Learning Engineer. 2024 to now.
- FedEx Ground, Data Scientist II. 2022 to 2024.
- FedEx Ground, Operations Research Engineer. 2021 to 2022.
- State Farm, Process Engineer. 2019 to 2020.
- Georgia Tech, MS in Computer Science (machine learning). 2020 to 2022.
- Arizona State, BS in Business Data Analytics. 2017 to 2019.
Mentoring and teaching
- Mentor to students at Georgia Tech and Arizona State since 2022, and a peer mentor on AI.
- Run ThredUp's ML intern program for 2025 and 2026. Hired three interns and a full-time engineer.
- Helped 8+ career changers build the skills to land data science roles.
- Taught a computer vision workshop for 15+ data scientists at FedEx.
Talks
- What is a machine learning engineer, and how to break into the field, Estrella Mountain Community College, Work Ready Wednesday
- Automating ML experiments with autoresearch, ThredUp AI engineering productivity series
- Running LLMs locally, ThredUp AI engineering productivity series
- A generative image explorer for product photos, ThredUp AI Showcase
- Flaw detection with computer vision, ThredUp operations update
- Flat-lay measurement R&D, Distribution center operations offsite
Writing
Advising
I take on a small number of advising engagements with operations and health tech teams that want ML they can rely on: loss prevention, delivery estimates, warehouse computer vision, or working out where to start. Past work includes analytics engineering and ML development for Float Health, an early-stage health care company, in 2024.
Get in touch
I like hearing about hard operations and data platform problems, leadership roles, and people getting started in ML. The fastest way to reach me is a message on LinkedIn.