Heather Dunlop Couture

About Heather Dunlop Couture

Most vision AI models work beautifully in the lab—until they meet the real world.In high-stakes domains like digital pathology and Earth observation, state-of-the-art test scores are meaningless if your model collapses when faced with a new scanner, a different geography, or sensor-specific noise.Providing Series A+ startups with late-stage technical authority, I embed with your team as a long-term partner to own scientific integrity and de-risk your roadmap. I focus on the technical hurdles that stall deployment and erode stakeholder trust:• : Ensuring models generalize across disparate sites and sensors.• & : Identifying and mitigating technical noise that masks true biological or environmental signals.• : Translating complex CV breakthroughs into peer-reviewed evidence and credible narratives for investors and regulators. With 20+ years of experience across 20+ organizations, I’ve identified the failure modes that stall vision AI—saving your team from learning the hard way.Recent work includes:• : Built the first independent, satellite-based system to monitor global power plant emissions (presented by Al Gore at COP26-30).• : Advanced a label-free platform for real-time cancer detection to an 82% sensitivity and 91% specificity, published in the Archives of Pathology & Laboratory Medicine?Founders and CTOs bring me in when they need a heavy hitter to:1: Move from research exploration to capital-efficient, robust product development.2: Level up junior ML engineers by instilling the technical rigor required for high-stakes AI.3: Diagnose why performance is failing in the field and pivot the technical approach before time and capital run out.If your vision AI needs to work outside the lab—and deliver impact across people or planetary health—let’s connect: https://pixelscientia.com/call/

Experience

1. Podcast Host

   Pixel Scientia Labs

   Oct 2022 — Present

   Learn how to build a mission-driven machine learning company from the innovators and entrepreneurs who are leading the way. A weekly show about the intersection of ML and business – particularly startups. We discuss the challenges and best practices for working with data, mitigating bias, dealing with regulatory processes, collaborating across disciplines, recruiting and onboarding, maximizing impact, and more.

Education

• Carnegie Mellon University

  Master of Science, Robotics

  2005 — 2006

• University of Waterloo

  Bachelor of Math, Computer Science

  2000 — 2005

• The University of North Carolina at Chapel Hill

  Doctor of Philosophy (Ph.D.), Computer Science

  2012 — 2018

Skills

• Opencv
• Gpu
• Pattern Recognition
• Image Processing
• Algorithms
• Medical Image Analysis
• Software Development
• Python
• Latex
• Parallel Computing
• Machine Vision
• C++
• Matlab
• Image Analysis
• C
• Programming
• Machine Learning
• Computer Vision
• Linux
• Deep Learning