Eric Wait, PhD

Technical Leader | Principal Data Scientist
Product Strategist

I bridge technical teams and product
vision to build systems that ship

With 21 years of military leadership, elite research experience, and startup product development, I bring a rare combination: deep technical expertise in GPU-accelerated ML systems and the ability to translate complex problems into executable strategies that align cross-functional teams.

What I Bring

🌉

Translation & Communication

Rare combination of deep technical expertise and exceptional people skills. I bridge technical teams and product management to accelerate development and improve team cohesion.

🚀

Product Vision to Reality

From "what if?" wonderment to deployed systems. I shape strategy, align stakeholders, and deliver production-ready solutions—90%+ reduction in analysis times, FDA-aligned systems.

Cross-Functional Leadership

21 years military leadership (2x Meritorious Service Medal). Communication under pressure, operational discipline, and building capability in others through empathetic mentorship.

How I Solve Problems

🎯 Outcome-Driven Discipline

I ensure teams build what's needed, not just what's possible. This means returning to "what does success look like?" at every decision point—when project pressures mount, when technically impressive alternatives emerge, when scope threatens to creep.

I've applied this discipline advising 170+ scientists, preventing wasted effort on experiments that couldn't answer their core questions.

Published framework: Journal of Cell Science (2020) - "Hypothesis-driven quantitative fluorescence microscopy"

🧠 Domain-Informed ML: "The Best Model is the One You Don't Need to Train"

By encoding physical constraints and domain knowledge directly into ML architectures, I build systems that are more efficient, interpretable, and robust than purely data-driven approaches.

80%
Less training data through physics priors
60%
Inference speedup via optical constraints
100x
Speedups with custom CUDA kernels

Delivering Impact at Scale

90%+
Faster Analysis

Led cross-functional teams to production clinical tools with dramatic efficiency gains and multi-site reproducibility

170+
Scientists Advised

Applied outcome-driven discipline enabling discoveries published in Nature and Nature Communications

12+
Years Building What Lasts

Open-source GPU libraries globally adopted—still actively used because thoughtful design serves real workflows

What Makes Me Different

The Rare Combination: Technical Depth + Exceptional Communication

Most technical leaders come from one of two paths: deep technical background with limited people skills, or strong communication skills with surface-level technical understanding.

I bring both: PhD-level technical expertise in GPU-accelerated ML systems AND exceptional ability to translate complexity into clarity. This combination accelerates development, improves team cohesion, and ensures technical excellence serves business goals.

Diverse Leadership Background

21 years military service (Air National Guard, 2x Meritorious Service Medal) taught me communication under pressure and operational discipline. Elite research institutions honed problem-solving methodology and advising scientists across domains. Startup product leadership delivered production systems balancing innovation with execution discipline.

Learning to See Without the Camera

Professional photography training taught me pattern recognition that now informs data visualization, compositional discipline that ensures reproducible systems, and a fundamental truth: the best solution often requires subtracting, not adding—whether in photographs or ML architectures.

"The camera is an instrument that teaches people how to see without a camera." — Dorothea Lange

Technical Approach: Physics-Informed ML

For those interested in the technical details of how I build efficient ML systems

More Efficient

Encode physical constraints to reduce training data needs by 80%

🔍

More Interpretable

Models that follow known physical laws are easier to debug and trust

🛡️

More Robust

Systems that respect physics generalize beyond training distribution

Concrete Examples:

  • Encoding optical constraints into imaging pipelines → 60% inference speedup
  • Using domain knowledge to reduce training data needs by 80%
  • Respecting physical limitations enables edge deployment
  • Custom CUDA kernels achieve 100x speedups over CPU implementations

Let's Build Something That Ships

If you need someone who can shape product vision, bridge technical and business teams, and deliver production ML systems—let's connect.