The Artificial Intelligence Podcast Interview #76 Zachary Hanif, VP of AI ML at Twilio
Sep 7, 2025
Zachary Hanif, VP of Data and AI at Twilio, brings a wealth of experience leading AI initiatives in both regulated financial services and communication platforms. He dives into the delicate balance between explainable AI and high-performing black box models, stressing the importance of tailored governance frameworks. Hanif discusses the challenges of moving AI from proof-of-concept to production, with 80% of pilots failing. He emphasizes the role of privacy-by-design principles and collaboration between tech teams and domain experts for successful implementations.
AI Snips
Chapters
Transcript
Episode notes
Design For Rigor From Day One
- Designing AI systems with auditability, predictability and robustness up front lowers maintenance costs and speeds delivery.
- Zachary Hanif says this rigor also improves onboarding and operational clarity across teams.
Pick Explainability Per Use Case
- Identify where each use case sits on the explainability spectrum before choosing model types.
- Avoid blanket explainability mandates that rule out high-performing blackbox models unnecessarily.
Enforce Data And Model Isolation
- Keep each customer's data and model instances isolated and let customers choose what to ingest into their private instance.
- This segregation preserves privacy while enabling personalized LLM behavior.