I move real business work onto AI, govern it, and prove it got better.
Twenty years leading software teams. Machine learning in production since 2016. Now running a global compliance program on AI agents. Six US patents.
Click anything. Drag or scroll to look around.
Have a seat
Rich Sadowski
I started in storage engineering, where automation cut outages that breached customer SLAs by more than 75% across 32 data centers. That set the pattern for the rest of my career: find the repeated work, automate it, and measure whether it got better.
At Alpine Access I ran software development for a platform that put 7,500 home-based agents to work, and co-invented a way to take card payments on a call without the agent ever hearing or seeing the number. At Foundever I led software R&D for more than ten years, with a team in eight countries, and co-invented the machine learning behind the OneTeam platform, used by 50,000 agents worldwide.
Now I manage compliance at Alchemer: SOC 2 Type 2, ISO 27001, GDPR, HIPAA, and the company’s Responsible AI program on ISO/IEC 42001. I moved most of the recurring work onto agents. The part I care about most is the part people skip: proving the automation still works and showing where it improved.
What people say
“Rich is a strategic leader whose big thinking and ‘leadership by example’ empower those around him to deliver. His keen insights and observations help him to dig into any complex topic and turn it into an actionable and attainable plan.”
Stephen Berdy · ThinScale Technology
“Rich has proven to be exemplary in his ability to communicate IT, infrastructure, telephony and security concepts in the company of both tech experts and non-technical participants.”
David Parkhurst · VXI Global Solutions
“Rich is very knowledgeable in all aspects of managing a software development project, along with a strong understanding of the technical aspects as well.”
Gary Castilla · Consultant to Foundever
Work samples
Real work, sanitized.
Company names, people and internal details are removed.
Don’t just automate the work, automate the improvement. Four review loops turn every run’s evidence into proposals a person approves, on a six-layer governed platform.
A compliance program run by agents, checked by people
Problem. One compliance manager, seven frameworks (SOC 2 Type 2, ISO 27001, GDPR, HIPAA, CCPA, FERPA, EU-US DPF), and hundreds of customer security assessments a year. Answering from memory is slow and drifts out of date.
Built a governed knowledge base, one maintained page per topic, so every answer traces to a source.
Agents draft assessment answers only where the knowledge base supports them and leave the rest blank. A separate read-only agent re-checks every flagged item. I review and submit.
Every corrected answer is logged with its cause, and the source is fixed so the error doesn’t come back.
Wrote the company’s Responsible AI program on ISO/IEC 42001: policy, risk and impact assessments, systems register, model cards. Eight documents, approved by the CISO.
Put the tooling under a CI gate of 2,600+ automated tests that blocks any change that fails, including my own.
Result. Customer security assessment turnaround went from 5 days to 2 hours, with every answer traced to a source and reviewed by a person before it goes out.
02 · Foundever (formerly SYKES) · 2012 – 2023
Machine learning that measured and improved customer service
Problem. A global customer-service company needed to know what was happening on every call and screen, at scale, without adding supervisors.
Led the R&D behind the OneTeam platform and co-invented five US patents behind it.
Built and ran a global team across eight countries.
Machine learning that labels customer interactions and predicts long or problem calls. Sentiment scoring on live call audio. Image recognition for workspace checks.
Security built in: agent login verified by hardware ID and location, and automatic masking of Social Security and account numbers in every event and recording.
Result. OneTeam ran for 50,000 agents worldwide, cut call handling time 23% and fraud 84%. Five US patents granted between 2020 and 2025.
03 · Alpine Access · 2007 – 2012
Secure payments for a work-from-home call center
Problem. 7,500+ agents worked from home. Taking credit card payments on calls meant card numbers in private homes, a PCI problem.
Co-invented SecureTalk (US Patent 9,100,484). When the caller keys in card digits, the system hides them from the agent’s ears and screen while the agent stays on the line to help.
No transfer to an automated phone menu.
Owned PCI, HIPAA, and SOX compliance for the platform.
Result. PCI protection for a fully remote workforce, while keeping a person on every call.
Typical turnaround
Security assessment turnaround
5 days
Every answer traced to a source and reviewed by a person before it goes out.
How it is governed
01 · KNOWLEDGE BASEOne maintained page per topic, each answer traced to a source.
02 · AGENT DRAFTSAnswers only what a source supports. The rest stays blank.
03 · CHECKER AGENTA separate read-only agent re-checks every flagged item.
04 · PERSON APPROVESI review every answer. Nothing ships unreviewed.
05 · CUSTOMERTurnaround: 5 days down to 2 hours.
↺ Every correction is logged with its cause, and the source page is fixed
Experience
Three drawers, four roles
2025 – now · Alchemer
Global Compliance Manager
Run the compliance program on agents I built and operate, with people approving every change. Security assessments went from 5 days to 2 hours.
2012 – 2023 · Foundever (formerly SYKES)
Global Executive Director, Software R&D
Led global software R&D with a team in eight countries. Built OneTeam, used by 50,000 agents worldwide. Co-inventor on five patents.
2007 – 2012 · Alpine Access
Vice President of Software Development
Built the SaaS platform for 7,500+ remote agents. Co-invented SecureTalk.
1999 – 2007 · Sungard Availability Services
Director of Storage Engineering
Storage engineering across 32 data centers. Automation cut outages that breached customer SLAs by more than 75%.
Education
Three degrees, one university: Stony Brook
1995 – 1996
MBA, Business (Honors)
1993 – 1995
MS, Materials Science & Engineering
Top of class.
1989 – 1993
BS, Materials Science & Engineering
Go Hawks
Super Bowl LX champions
29 – 13
Seattle over New England, February 8, 2026, at Levi’s Stadium in Santa Clara. The defense held the Patriots scoreless for three quarters, and Kenneth Walker was named MVP.
Still the best thing on this wall that I didn’t patent.
“Suddenly, our computers and all our connected devices will be full of algorithms that, like weeds, will adapt and move swiftly, seeking new ways to harvest energy and capture living space.”
Joelle Jenny, Harvard Weatherhead Center for International Affairs
“So we must assume that crises and severe disruptions will happen, and work toward making our digital infrastructure less vulnerable on the one hand, and better able to recover from crises on the other hand.”
Joelle Jenny, Harvard Weatherhead Center for International Affairs
Competing in the Age of AI, Marco Iansiti and Karim Lakhani
Prediction Machines, Ajay Agrawal, Joshua Gans and Avi Goldfarb
The Coming Wave, Mustafa Suleyman
Human Compatible, Stuart Russell
The Alignment Problem, Brian Christian
Superagency, Reid Hoffman and Greg Beato
Genius Makers, Cade Metz
AI Snake Oil, Arvind Narayanan and Sayash Kapoor
AI Superpowers, Kai-Fu Lee
Life 3.0, Max Tegmark
Power and Prediction, Ajay Agrawal, Joshua Gans and Avi Goldfarb
Nexus, Yuval Noah Harari
The Worlds I See, Fei-Fei Li
Second Brain
Five lessons from running a second brain
My compliance knowledge base is an LLM-maintained wiki kept as an Obsidian ↗ vault. Source documents go in, an agent writes and maintains the pages, and every answer cites the page it came from. These five practices held up in daily use.
Keep sources and knowledge apart. Raw documents are read-only. The agent owns the wiki pages built on top of them, so no original is ever lost and any page can be rebuilt.
One source of truth, no conflict list. When two pages disagree, fix it on the spot. A real unknown is marked OPEN on the page where someone would act on it, never in a separate list.
The newer fact wins. Date every fact and ruling. When sources disagree, take the newer one and fix every page that repeats the old one. A standing ruling holds until its owner changes it.
Every correction goes back the same day. Fixing only the deliverable does not count. If the page stays wrong, the same mistake comes back on the next answer.
Enforce the rules with code, not prose. A script blocks any answer that does not cite a real page and an exact quote. Pulling the right sections before whole pages cut the text read per answer by 47%.
Tool
Obsidian
Plain Markdown files on my own disk, with links and a graph view to see which pages carry the most answers.
Out the back: the foothills and a park. Out the front: the street, where the neighbors walk their dogs. The sky follows your own time of day. The wall clock keeps Denver time.