Michael Kaminski — I build AI agents that run in production
Agent infrastructure, MCP servers, and eval harnesses inside a regulated lender.
Python and TypeScript. Atlanta, relocating to New York City.
What I'm looking for: senior product roles at the agent layer —
Technical Product Manager, Senior Product Owner, or PM for an agent platform — at teams
shipping AI agents into production, especially where the environment is regulated and the
evidence trail matters. Open to fractional and full-time.
I'm a technical product manager working at the agent layer. I took an AI agent
capability from prototype through security, legal, and compliance review to production
inside a regulated lender, and I build custom Model Context Protocol (MCP) servers,
multi-agent pipelines, and eval harnesses in Python and TypeScript. What is unusual about
the combination is that I can price a system and I can ship it: before the agent work I
spent years in corporate finance and strategy, and I have authored Terraform modules with
multi-environment remote state and operated production Kubernetes clusters.
Latest:Nvidia + Hugging Face: $12.9B to Close the Gap Between a Model Card and a Running Kernel — Nvidia reportedly agreed to buy Hugging Face for $12.9 billion — nine months after Hugging Face turned down a $500M investment at a $7B valuation, an 84% step-up. I am long NVDA and Hugging Face is where my weights come from, so read the bull case accordingly. The honest arithmetic: my break-even for self-hosting is 3.04 billion tokens a month and I am nowhere near it. The line item this deal actually attacks is not the GPU-hour. It is the week of engineering between a download and a running kernel.
About
I build AI agents that run in production. Most of what that involves is not the model — it is the tool surface the model is allowed to call, the evidence the system has to produce afterwards, and the gates that stop it doing anything it cannot undo.
I came to it sideways. I spent years in corporate finance and strategy — capital markets, divestitures, an IPO, a $4B share repurchase — and the through-line is the same in both halves of that career: the number has to reconcile, and somebody has to be able to check it eighteen months later. That instinct turns out to be exactly what an agent needs before it can go anywhere near a regulated decision.
So I can price a system and I can ship it. I have authored Terraform modules with multi-environment remote state and run production Kubernetes; I have also built the financial model a private-equity board expected. Very few people building agent infrastructure can also make the business case for it, and that combination is the whole of what I am useful for.
I write about the parts that went wrong, because those are the parts worth reading. The statistical gate I built for my own agent's instruction file turned out to be decorative once I ran the power calculation on it. A pipeline I looked at daily was quietly rewriting its own history until I added one more file. Both are on this site with the arithmetic shown.
Took an AI agent capability from prototype through security, legal, and compliance review into production at a regulated consumer lender.
My role: I owned the agent capability end to end: the architecture, the tool surface the model was allowed to call, the evidence design that got it through legal and compliance review, and the approval gates on irreversible actions. Engineering built alongside me; the review strategy, the retention decision, and the control design were mine.
A build that publishes 1,180 records four ways — static pages, an interactive graph, a search index, and an MCP server — in 0.39 seconds with zero dependencies.
My role: I designed and built the whole pipeline: the source ontology, the build, all four output surfaces, and the MCP tool design. The decision worth copying — having the MCP server import a build artifact rather than query the site — is the one I would defend hardest.
A dealer-gamma pipeline that recomputes option-chain levels every fifteen minutes — and the append-only write that caught it moving levels on a closed market.
My role: I built the pipeline and the study it writes, and I found the defect. The fix was not a code change so much as an instrumentation decision: the system was already wrong, and the only reason it became visible is that I stopped letting it overwrite its own evidence.
Stellantis Financial Services —
Senior Product Owner ·
Product ownership for consumer auto-finance platforms — servicing, payments, and lending systems. The AI agent work sits inside this role: agentic workflow design, custom MCP servers, eval harnesses, and human approval gates on irreversible actions, taken from prototype through security, legal, and compliance review into production. Current role
Fyxed —
Interim CEO ·
Interim CEO of Fyxed, a fintech company that provides a platform for small businesses to manage their finances. Established GTM strategy, built product, hired team, and scaled to $50k MRR
Momnt —
Senior Manager, Product Engineering ·
Fintech platform development and credit reporting systems Launched credit reporting, accounting engine product owner, departed after founders left - culture change
Property Walk —
Founder ·
Founder of Property Walk, a real estate technology company that provides a platform for property managers to reduce service calls and overbillings of residential home services. 5k MRR, PMF not found
Superior Contracting & Maintenance —
Co-Founder ·
Co-founder of Superior Contracting & Maintenance, a construction company that provides services to the residential and commercial sectors. Installed CEO & NetSuite ERP, left to focus on high-growth fintech
GreenSky —
Product Manager, Credit & Strategy ·
Product management and strategy for the GreenSky credit platform Post IPO, left to found Residential Services company with brother
HD Supply —
Senior Analyst, Strategic Finance ·
Divestiture transactions and operational transformation Post 2,500 FTE restructuring, changed focus to grow revenue, rather than enhance margins through OpEx cost out
KPMG —
Senior Consultant, Advisory Services ·
Dispute advisory services, tax structuring, & data analytics Found wife in consulting, left for M&A work @ HD Supply
ModularEquity / MEK Capital —
Founder, Managing Partner ·
Long-Only diversified investments across PE, Real Estate, Equities, & Debt, managing $3mm in capital Active Management
Home Depot —
Senior Analyst, Merchandising Finance & Treasury ·
Share repurchase programs and financial strategy Departed to obtian diversified experiences within other companies/industries
Capital markets and corporate development work from the finance half of the career.
Every figure below is a single named transaction; the total is their sum, not an estimate.
Date
Company
Type
Class
Value
Counterparty
Jun 2024
Momnt
144(a)
Debt
$400M
Saluda Grade
Sep 2023
Momnt
Series B
Equity
$15M
TruStage Ventures
May 2018
GreenSky
S1
Equity
$1,010M
Public Markets
Dec 2017
GreenSky
Debt Facility
Debt
$200M
PIMCO
Apr 2016
HD Supply
Senior Unsecured
Debt
$1,000M
Public Markets
Feb 2015
HD Supply
Divestiture
Equity
$825M
Power Solutions
Apr 2014
HD Supply
Divestiture
Equity
$90M
Crown Bolt
Apr 2014
HD Supply
Secondary
Equity
$957M
Bain, Carlyle, Dublier & Rice
May 2012
Home Depot
Share Repurchase
Equity
$4,000M
Goldman
Oct 2011
Home Depot
Line of Credit
Debt
$700M
Citi
Sep 2011
Home Depot
Senior Unsecured
Debt
$2,000M
Public Markets
Total
$11,197M
Writing
Nvidia + Hugging Face: $12.9B to Close the Gap Between a Model Card and a Running Kernel — Nvidia reportedly agreed to buy Hugging Face for $12.9 billion — nine months after Hugging Face turned down a $500M investment at a $7B valuation, an 84% step-up. I am long NVDA and Hugging Face is where my weights come from, so read the bull case accordingly. The honest arithmetic: my break-even for self-hosting is 3.04 billion tokens a month and I am nowhere near it. The line item this deal actually attacks is not the GPU-hour. It is the week of engineering between a download and a running kernel.
One Knowledge Base, Four Surfaces: Pages, Graph, Search Index, and MCP — The Genome of Games publishes 1,180 records four ways — 1,245 static pages, an interactive graph, a 126 KB search index, and an 8-tool MCP server. One build writes all four in 0.39 seconds. The MCP server never queries the site; it imports a build artifact, so an agent and a crawler cannot disagree. The one surface the build does not own has already drifted by 7,492 links.
Shipping an AI Agent Through Compliance Review in Regulated Lending — On May 12, 2025 the CFPB withdrew 67 guidance documents in a single notice, including both circulars covering adverse-action notices for complex algorithms. The underlying duty did not move. Build the agent against the statute and against the evidence it will have to produce — starting with a 120-day retention floor that most agent logs miss.
Statistical Gating for Agent Instruction Changes — Every edit to an agent's instruction file is an experiment. Mine requires a statistically significant improvement — Welch's t-test, p < 0.10, at least a 5% lift — against a 14-day rolling baseline before the change is allowed to stay. Then the power calculation showed the 14-day window can only detect a 0.97 standard-deviation shift, which makes the 5% threshold decorative.
I Gave My Website a Voice, and the Voice Was the Easy Part — There is now a digital twin on my homepage you can talk to out loud, hands-free, in a clone of my voice. Getting it to sound like me took an afternoon. Getting it to sound like a person took rewriting how it writes.
I Shipped the Same Pull Request Twice and Only Noticed the Second Time — Behind the Build, Vol. 8: two new tabs — Websites and Products — landed a day apart and turned out to be the exact same diff wearing different screenshots. What that repetition taught me about when a pattern is worth abstracting (and when it very much is not).
My Automations Didn't Crash. They Just Stopped Talking. — Behind the Build, Vol. 6: the daily-article robot quietly skipped four days without erroring once. A look at the fleet of small agents I now run across a portfolio site, a trading account, a resale business, and a spend audit — and why silence is the failure mode nobody instruments for.
I Built Analytics to Watch My Site, and Discovered It Was Watching Nothing — Behind the Build, Vol. 4: wiring up real product analytics turned up a silent reverse proxy eating every POST request, a dropped entry pageview, and a Gantt chart bug where two bars collided despite their dates never touching.
I Got Locked Out of My Own GitHub Account (Sort Of), So I Found the Back Door — Behind the Build, Vol. 3: adding a Products tab for my desktop tools meant fetching screenshots from repos my own session wasn't allowed to touch — and learning the honest way to handle a missing screenshot instead of faking one.
Why Fintech Belongs in Atlanta — and Why That Matters for Your Cap Table — Atlanta quietly processes a huge share of the world's card transactions and has produced a run of fintech exits. Here's what that ecosystem means for founders building here — and the finance-plus-engineering talent it creates.
Quality of Earnings, Explained for Fintech Founders — What a Quality of Earnings (QoE) analysis actually looks at, why fintech QoE is different, and how founders can be ready before a buyer or lender runs one.
I had the pleasure of working with Michael, and I can confidently say he is an outstanding Product Engineer. He has exceptional communication skills and a unique ability to break down complex technical challenges into clear, actionable solutions, making collaboration seamless across teams.
What truly makes Michael stand out is his problem solving mindset. He doesn't just fix issues; he digs deep to identify root causes and proactively implements long-term solutions that improve both product performance and processes. His understanding of regulatory compliance is particularly impressive. He recognizes its critical role in product development and ensures that every solution aligns with industry standards.
Beyond his technical expertise, Michael is also highly skilled in documenting processes and simplifying complex systems so that both technical and non-technical users can understand them. He has a knack for translating intricate engineering concepts into clear, accessible documentation that enhances efficiency, knowledge sharing, and overall user experience.
Michael's expertise, attention to detail, and commitment to excellence make him an invaluable asset to any team. I highly recommend him to anyone looking for a skilled and strategic engineer who drives meaningful impact.
I've had the privilege of working with Michael at Momnt, and he is truly one of the most capable and driven professionals I've encountered. His ability to optimize complex data pipelines, streamline operations, and align technology with business goals is outstanding. Whether tackling intricate technical challenges or supporting colleagues with thoughtful insights, Michael consistently goes above and beyond.
Beyond his technical expertise in SQL, Python, and data engineering, his collaborative spirit and problem-solving mindset are what sets Michael apart. He's always the first to jump in when help is needed, ensuring that both strategic and operational goals are met efficiently. His leadership and analytical skills have directly contributed to business success, and his positive attitude makes him a pleasure to work with.
Any organization looking for a highly skilled, adaptable, and dedicated professional would be lucky to have Michael on their team. I highly recommend him for any opportunity that requires both deep technical acumen and strategic vision.
I had the pleasure of working with Michael ("Kaminski") at Momnt, where he played a crucial role in managing complex data, optimizing pipelines, and ensuring seamless operations. Beyond his technical expertise, Michael is a great collaborator who brings a positive attitude and problem-solving mindset to every project. He's always appreciative of others and a fantastic team player. Any organization would be lucky to have him!
Abby Martin — Senior Product Manager, Momnt · LinkedIn
I worked very closely with Michael during his time as CFO for Superior Contracting & Maintenance and would not hesitate to recommend him for Senior and Executive leadership roles.
Michael's superpower is a double whammy of emotional intelligence and strategic vision. A true servant leader that endeavors to learn and understand his team and employees and their motivators/drivers as part of the business, Michael places a premium on a positive culture to drive results from a happy, encouraged, and supported team. With a strategic vision and passion for technology, Michael brings a unique touch to his roles, often finding new tech and applications to streamline operational workflows before most people even become aware of the platform(s).
Michael is a perennial educator and facilitator. Over the years, I have watched Michael teach corporate financing and business to dozens of coworkers (myself included) simply because they asked and he truly believes education, personal/professional development, and understanding the "why" build great leaders and greater companies.
Michael is a first round draft pick and will be an outstanding impact player and leader for any company lucky enough to have him.
EJ Shramek — Director of Project Management, Superior Contracting & Maintenance · LinkedIn
In my professional career, I have never come across someone as willing & able to use their knowledge and technical skill set to advance the company as a whole as Michael.
In our time together at Momnt, he consistently went above and beyond to help others solve complex problems. When coworkers would post questions on our internal messaging system, Michael was usually the first to reach out with answers. And if he didn't know at the time, he would research the issue and come back with solutions - all while outperforming in his own job duties.
What most impressed me about Michael was his drive. It became clear early on that he was not motivated by public accolades or promotions - he simply wanted to do the best job possible at any task in front of him for the benefit of the company.
Any business looking for someone highly skilled in coding (SQL, Python,etc.), statistics, data analysis, financial modeling, strategy, and countless other hard & soft skills would greatly benefit from having Michael on staff.
I had the privilege of working closely with Michael Kaminski during their time as CFO at Superior Contracting & Maintenance, and I can confidently say that they are one of the most talented and versatile leaders I've had the pleasure of working with.
Michael carries a deep acumen for finance with a forward-thinking approach to technology and innovation. His leadership extended well beyond traditional financial management playing a pivotal role in aligning financial goals with technology investments, streamlining operations, and guiding the organization through complex transformations.
Michael is a natural problem solver, with a keen understanding of how to leverage technology to optimize business processes and drive efficiency, willing to do so without leaving any option undiscovered and placing a one of a kind drive into discovering solutions. Whether it was overseeing large-scale system upgrades or aligning financial and technological strategies, Michael shows a remarkable aptitude for organization and leadership in any capacity.
Michael has a unique talent of meshing a 'Big Picture' mindset while maintaining a true genuine, compassionate approach to management.
I have no doubt that Michael would excel in any senior leadership role.
Daniel Walker — Director of Construction, Superior Contracting & Maintenance · LinkedIn
I had the pleasure of working alongside Kaminski, and I can confidently say he is one of the most intelligent and results-driven professionals I've encountered. His ability to engineer high-performance data solutions and drive efficiency at scale is truly impressive.
His leadership in managing a development team while directly supporting executive leadership showcased his ability to balance both strategic and technical initiatives seamlessly. His efforts directly lead to tens of millions in capital being generated or saved that otherwise would have never come to fruition.
One of his standout achievements was leading critical reporting and credit functions for Capital Markets transactions, facilitating deals exceeding $300M in unsecured loans. His expertise in Python, SQL, AWS, and data engineering was instrumental in enhancing risk, compliance, and operational workflows.
Beyond his technical expertise, Kaminski is a problem solver who thrives in high-stakes environments. His contributions in integrating external risk datasets and engineering scalable infrastructure resulted in millions in recovered revenue and improved business intelligence. Any organization would be fortunate to have him on their team, and I highly recommend him for any role that requires a blend of strategic thinking and deep technical acumen.
Michael helped our group every time we needed him. He is a great communicator and handles multiple projects while always makes room for others. I recommend him highly and it was such a pleasure to work with him.
I've had the pleasure of working alongside Michael, and I can confidently say his ability to align technology with business strategy is truly impressive, consistently driving innovation and efficiency. Any organization would be fortunate to have him.
I had the pleasure of working with Michael Kaminski on various levels during my graduate career at Georgia State University. One was a highly complex project for the University's Portfolio Management Team. Michael, the President of the team, was a highly effective leader that received the best work from each individual team member. It was his leadership skills that ultimately drove the team to success.
I recommend Michael Kaminski for any leadership role in a financial setting.
Michael Hill — Manager, GPS EFA, Deloitte · LinkedIn
Michael is an excellent and organized Chief Investment Officer of Robinson's student managed equity portfolio. The portfolio is doing extremely well and the team is working together as a cohesive group. That's what the CIO is supposed to do and he is executing extremely well.