Career Experience

Where I've been, what I built.

2021–August 2026

ML Eng + Technical Product Manager, Equilibrium Energy

Built the Science platform that took the company from Seed to Series B. Experiments, deployment, observability, and agentic AI for forecasting and decision algorithms.

2017–2021

ML Platform Engineer, Optum (UHG)

Built everything needed to support ML lifecycle from experiment -> production. IaC, CI/CD, SWE, MLOps, testing, validation, and monitoring.

2017

B.S. Computer Science, University of Minnesota

Computer science because the problem solving is fun and because it can help solve problems in any domain.

Roman Grebin

Minneapolis · Eng · Technical Product Manager · Product Engineer

Learning, growing, and mission-oriented problem-solving. Resting and exploring on “sabbatical” while deciding what to do next. Product engineering with Buy Nothing Project, Twin Cities Climate Collaborative, and others.

Career Directions

Problems and directions that feel worth staying with.

Technology and social good

How can I leverage my skills to make the world a better place to live for myself, others, and future generations?

Getting closer to tangible impact

Do something that unambiguously improves lives.

Important problems at the edge of solvable

I'm drawn to the intersection of emerging tools, under-appreciated problems, and difficult domains. Early enough that the path isn't obvious, late enough that it's actually possible.

What has shaped me?

Places and Activities.

Where I've lived

In the Midwest. Chicago area until age 8. Dunbar, WI through high school. HS graduating class size: 16. Minneapolis ever since.

Sports

Various sports. Basketball was my favorite. Lots of Ultimate Frisbee in college. Bouldering is the new favorite. We built a bouldering wall in our house, frequently visible behind me in video calls.

Volunteering

Various organizations over the years (some listed on my résumé). Improving places we live, giving back to community, meeting good people, trying to find activities that energize me.

What I am up to.

Outside of work.

Exploring civic, community, and social tech

How can tech make our lives better and more connected, rather than replacing human connection? Made geographic.community, ourblock.community, and this personal website for fun, to learn E2E website dev, and to explore patterns for social tech.

Monthly checklist

Each month my wife and I "perfect" a recipe, engage with local community, do something stimulating, and engage in some sort of service work.

Enjoying the Cities

We love Minneapolis more than ever. Biking, walking, food, parks, neighborism, ...

Plans for the future

Things I'm looking forward to.

Enjoying outdoor activities

Biking, walking, some basketball this summer.

Community engagement

How do we show up better for neighbors? Show up in-person. Explore whether tech can help (see: Relational Tech).

Past
Present
Future
Career
Personal
View
Roman Grebin
ML Engineer · ML Platform Architect · Technical Product Manager
Experience
Equilibrium Energy
ML/AI Engineer and Technical Product Manager
July 2021–Present
  • Architected and built Equilibrium's internal Science Development Platform, scaling model experiments from hundreds to millions monthly. Platform underpinned operational trading and grid asset optimization, the core revenue-generating workflow powering us from Seed through Series B.
  • Designed and launched an analytics observability platform providing proactive and real-time performance visibility, enabling human-in-the-loop augmentation of automated decisions.
  • Partnered with data engineers, domain experts, and data scientists to develop and deploy machine learning models to forecast weather and energy prices on 5-minute, hourly, and daily intervals.
  • Pioneered shared agentic AI coding standards, and extended our agentic productivity framework to non-technical users with a self-service skill and plug-in repository marketplace synced to internal agents.
Optum (UHG)
Data Scientist and Machine Learning Engineer
May 2017–July 2021
  • Feature engineering, XGBoost modeling, and deployment of ML system to automate 25M+ identity matching tasks.
  • Defined deployments of Microsoft Azure virtual machines and cloud tools in Infrastructure as Code to enable 100+ data scientists while meeting PHI and HITRUST requirements.
  • Developed frameworks and automation for model evaluation, data visualization, and deployment.
  • Built algorithms to predict anomalous cost events using regression and gradient-boosted trees.
  • Wrote and installed Helm charts to deploy software including Airflow and MLFlow to Kubernetes.
Education
B.S. Computer Science
University of Minnesota
May 2017
Skills
AI Empowered Workflows — Leveraging skills, rules, and MCP services with primary commercial agents.
Cloud Development and Architecture — Azure and AWS via Terraform, Helm, Jenkins, Databricks, Docker, HDInsight, Kubernetes, Lambda, S3, and more.
Data Engineering — Airflow, Dagster, DVC, Great Expectations, Pandera, Spark, SQL, and more.
MLOps Tooling — AzureML, Feast, Kubeflow, Metaflow, MLFlow, SageMaker, Seldon, Weights & Biases, and more.
Python — Dash, pandas, PySpark, scikit-learn, Streamlit, XGBoost, and more.
In My Own Words
What I Bring

An engineer who thinks like a product manager, and a product manager who understands technical systems from top-to-bottom and can build independently. I learn fast, build from 0->1, and have a calibrated sense for when a problem calls for speed and when it calls for rigor.

What I Build

I build systems that empower others to do better work. At their best, platforms disappear into the background: scientists run experiments without thinking about infrastructure, operators get alerts before things go wrong, teams collaborate without friction.

How I Think

I've lived and worked across many boundaries: engineering and product; healthcare and energy; startups and massive orgs. As a result, I know that the best solutions usually live where disciplines overlap. I'm most useful when the problem isn't fully defined yet, when it takes someone who can understand it, make decisions, build a solution, and see how it fits the larger picture.

What's Next

I'm drawn to problems that aren't getting the solutions they deserve. This is frequently at the intersection of:

  • The cutting edge: technologies that work in theory but haven't been cracked at scale
  • Hard problems: questions where the answer genuinely doesn't exist yet

This is exemplified by my work on cloud, ML, and agentic AI technologies in healthcare and energy. I want to keep building platforms, shaping teams, and maintaining a point of view across code, users, products, and how the system works.