Seyi Feyisetan
AI, privacy, and trust.
I am a Principal Scientist at Amazon working on LLMs, where I lead agentic initiatives
and science for Alexa+ capabilities like explainability and visual intelligence. I also
built anonymization techniques for Alexa, Rufus, Amazon Nova, and speech LLMs. Along the
way I have advised close to a hundred GenAI product initiatives across AWS, Alexa,
AGI Nova, Ring, and Fire TV on privacy, safety, personalization, and red teaming.
Before Amazon, I led research at Meta on differential privacy and secure multiparty
computation. I hold a PhD in Computer Science from the University of Southampton, I've
been granted multiple US patents, and my privacy work has shipped in large-scale
production systems. I founded the PrivateNLP workshop series and previously served on
the research advisory board of the IAPP.
These days I'm building something new.
Highlights
2026
- Part of the launch of Alexa for Shopping
- Launching visual intelligence on new AI devices
- Building new trust and safety frameworks for agentic systems
2025
- Led science for Alexa+ Explainability (“Alexa, why did you do that?”)
- Led science for Alexa+ Vision (“Alexa, what am I holding?”)
- Built anonymization techniques for Rufus, Amazon Nova, and speech LLMs
- Advised 40+ product initiatives across Alexa, Nova, Ring, and Fire TV, from scaling AI red teaming to guardrails for personalization and edge AI models
2024
- Technical reviewer for 45+ GenAI programs across Amazon, including LLM runtime security (PIN codes, security-critical APIs), automatic red teaming, and new end-to-end encryption designs for consumer devices
- Developed contextual integrity approaches for data transfer in LLM systems
- Partnered with legal on third-party LLM use and controls
2023
- Wrote the foreword for Ken Huang's book on LLM security
- Granted a US patent on information uniqueness determination
- Invited talks at Carnegie Mellon and the Amazon Data Conference; tutorial on privacy-preserving NLP at EACL
- PrivateNLP, the workshop series I founded, accepted at ACL 2024
2022
- Returned to Amazon as Principal Applied Scientist, Trust & Privacy
- Granted US patents on calibrated noise for text modification and data-preserving text redaction
- Keynote at Amazon's internal Privacy-Preserving Machine Learning workshop
- Led a session at the joint differential privacy workshop hosted by Google and Meta
2021
- Joined Meta to lead research on differential privacy and secure multiparty computation
- Published at ICML 2021 on label inference attacks; best paper award at FLAIRS 2021
- Granted a US patent on privacy- and intent-preserving redaction for text utterances
- Panelist at NeurIPS PriML alongside Helen Nissenbaum and Aaron Roth
- Appointed to the research advisory board of the IAPP
2020
- Founded the PrivateNLP workshop series (WSDM 2020, then EMNLP, NAACL, and ACL)
- Shipped metric differential privacy mechanisms for text into production systems at Amazon
- Invited talks at the University of Oxford and the University of Maine
Full publication list on
Google Scholar.
Contact
Email: seyi dot feyisetan at gmail
LinkedIn: in/SeyiFeyisetan