Senior Director of Analytics · Certified Chief Data Officer
I lead enterprise data and AI for a national healthcare organization, and I own both halves of the AI problem: the governance framework and the business enablement push. Guardrails and adoption come from the same desk, which is the only arrangement I have seen actually work.
I keep this profile because I think data and AI leaders should work the way engineering teams work. Version control, review, reproducibility. A governance policy that cannot be diffed is just an opinion, and a metric nobody can trace back to a query is just a number on a slide.
Working on: enterprise AI governance and responsible adoption · AI business enablement past the pilot stage · modern data platform and lakehouse architecture · self-service BI that people actually use · data as a product, with customers and revenue attached
The reference implementations and standards I read when building governance frameworks. Other people's code, listed because the thinking is worth borrowing.
| AIF360 |  |
IBM's fairness metrics and bias-mitigation algorithms for datasets and models |
| fairlearn |  |
Microsoft's toolkit for assessing and improving model fairness |
| AIX360 |  |
Interpretability and explainability across the model lifecycle |
| responsible-ai-toolbox |  |
Model and data assessment interfaces for responsible AI development |
| what-if-tool |  |
Google PAIR's no-code probing of model behavior and fairness |
| AI Incident Database |  |
Cataloged record of real-world AI harms. The failure library governance is written against |
| AwesomeResponsibleAI |  |
Curated research, regulations, maturity models, and standards |
| ai-governance-framework |  |
FINOS framework for governing AI in regulated financial services |
| fabric-toolbox |  |
Accelerators and patterns for Microsoft Fabric, from the Fabric CAT team |
| Data-and-Agent-Governance-Accelerator |  |
End-to-end AI governance across Purview, Copilot, Fabric, and custom agents |
| purview-data-governance-masterclass |  |
An opinionated end-to-end data governance implementation |
| great_expectations |  |
Data quality validation. The contract layer under any trustworthy dashboard |
| OpenLineage |  |
Open standard for lineage metadata. Provenance you can query |
| mlflow |  |
Tracking, evaluation, and deployment for ML and agent systems |
| awesome-data-engineering |  |
Curated data engineering tooling |
| awesome-mlops |  |
Curated MLOps tools and platforms |
| GoogleCloudPlatform/generative-ai |  |
Enterprise GenAI patterns and notebooks on Google Cloud |
| gemini cookbook |  |
Examples and guides for the Gemini API |
| claude-cookbooks |  |
Anthropic's recipes for building effectively with Claude |
| claude-code |  |
Agentic coding in the terminal. How this profile gets written and versioned |
| how-to-build-a-coding-agent |  |
Building a coding agent from scratch, to understand what is under the abstraction |
| aws-well-architected-labs |  |
Hands-on labs for architectural best practice |
Certifications — Chief Data Officer, Carnegie Mellon (2023) · Google Cloud Digital Leader (2024) · Generative AI, LLMs & Responsible AI, Google Cloud (2024) · Google Cloud Professional Cloud Architect (in progress) · Healthcare Management Professional, AHIP (2012)
Thought leadership — CDOIQ Review Board, MIT International Conference on Information Quality (2024) · Featured panelist, CDO Magazine Data & Security Summit Wisconsin (2024)
Connect — linkedin.com/in/irynafeldman