Find Your Model's Blind Spot in 60 Minutes
Join Nahid Farady, PhD (Principal Tech Lead, AI Security & Privacy · Microsoft) for a free live session.
Nahid Farady, PhD
Principal Tech Lead, AI Security & Privacy · Microsoft
⭐ 4.9 / 5
You shipped the model. Did you audit it?
ML engineers and data scientists ship AI systems that make real decisions about real people — and most have never run a formal fairness or explainability audit on what they've built. This course gives you the tools, frameworks, and documentation to run a responsible AI audit on a live or provided system — bias detection with AI Fairness 360, explainability with SHAP and LIME, and an accountability report your legal and compliance team can act on.
What You'll Learn
Bias Audit Report
Scored fairness audit using AI Fairness 360 — metric selected against IEEE P7003 guidance (disparate impact, equalized odds, statistical parity, or individual fairness), not just whichever one runs first. Findings scored across demographic groups with remediation prioritized by impact.
Explainability Integration
SHAP for global feature importance and individual prediction explanations, LIME for model-agnostic local approximations — with documentation of where each tool's assumptions hold and where they break down. Extended to LLM outputs with tiered documentation for engineering, legal, and end users.
Accountability and Model Card
A model card built to EU AI Act transparency requirements — documenting intended use, out-of-scope uses, known limitations, bias findings, human oversight requirements, and update/deprecation policy. Structured for technical and legal audiences.
Full Responsible AI Audit Report
Bias audit, remediation report, explainability documentation, and model card in a single audit package — with an executive summary that maps findings to decisions without requiring stakeholders to read the underlying technical docs.
Who Is This For?
This course is designed for:
ML Engineers Who've Never Audited What They've Shipped
Who build and deploy models that make consequential decisions and know fairness and explainability audits are overdue — but have never had a structured process for running one.
Data Scientists Asked to Explain Model Decisions to Non-Technical Stakeholders
Who are being asked by product, legal, or leadership to justify model outputs and need tooling and documentation frameworks to do it credibly.
AI Product Managers Responsible for Model Accountability
Who own AI products that touch users directly and need to implement responsible AI practices before a regulatory requirement, audit, or public incident forces the conversation.
Course Outline
4 weeks · 3 sessions per week
Projects You'll Ship
Leave with real work to show, not just a certificate.
Bias Audit + Remediation Report
Fairness audit using AI Fairness 360 — metrics selected against IEEE P7003 guidance, findings scored across demographic groups, and remediation applied with explicit tradeoff documentation and residual risk rationale. Reusable as a repeatable template for every model release.
Explainability Documentation Package
SHAP and LIME on a real or provided model with tiered documentation for engineering, legal, and end-user audiences. Includes LLM and generative AI explainability coverage and EU AI Act transparency alignment notes.
Full Responsible AI Audit Report
Bias audit, remediation decisions, explainability documentation, and model card in a single package — with an executive summary and EU AI Act transparency compliance notes. Structured for compliance review, public disclosure, and repeatable use.
Your Instructors
Nahid Farady, PhD
Principal Tech Lead, AI Security & Privacy · Microsoft
⭐ 4.9 / 5
Nahid leads AI security, privacy, and responsible AI engineering at Microsoft Copilot, with prior roles at Google Cloud and Capital One CyberML. She holds a PhD from Virginia Tech and brings 10+ years of applied experience in cybersecurity, threat modeling, and ML deployment at scale. She also teaches AI and security as adjunct faculty at UC Berkeley.
What Students Say
⭐⭐⭐⭐⭐
"Running AI Fairness 360 on our fraud model in Week 1 surfaced a demographic disparity we'd had in production for eight months. We fixed it before it became a regulatory issue."
Jordan Park
ML Engineer · Brex
⭐⭐⭐⭐⭐
"The SHAP documentation package I built in Week 2 is now our standard for every model deployment. Our legal team stopped asking for ad-hoc explanations overnight."
Taylor Nguyen
Data Scientist · Lattice
⭐⭐⭐⭐⭐
"I shipped our model card and full audit report six weeks after the sprint. First time we had responsible AI documentation that passed legal review on the first submission."
Casey Kim
AI Product Manager · Rippling
Course Schedule
All sessions are instructor-led and live. Recordings available within 24 hours.
SUNDAY
9:00 AM PDT
Live ClassBias detection, explainability, and accountability — applied to real models in production.
WEDNESDAY
6:00 PM PDT
Lab SessionHands-on tooling work with AI Fairness 360, SHAP, and LIME on real or provided models.
THURSDAY
6:00 PM PDT
Build & ShipFinalize weekly deliverables with peer review and instructor calibration.
Frequently Asked Questions
Related reading
The JD-backed research behind this course — from Dexity Intel.
AI Governance in 2026: How to Govern the AI You Ship →How far do you want to go?
Start free to experience our offering, choose the program length you would want to commit to.
You build. Nobody demos at you.
Every session is follow-along — you build the thing yourself while a practitioner works beside you. That is why the hours look long: they are yours to build in, with an expert on hand to guide you. None of it is a traditional lecture.
In the session, you'll: