UC San Francisco
San Francisco, CA
Research Intern
2012
01 —
Navdeep Gill works at the intersection of Responsible AI, AI Governance, and AI Risk Management, developing strategies and solutions that ensure AI systems meet the highest standards of transparency, fairness, accountability, and robustness. Currently at ServiceNow, he collaborates with cross-functional teams to operationalize AI policies and risk frameworks, helping scale responsible AI practices across enterprise applications.
Good AI products come from tight collaboration between product, engineering, and research.
During his time at H2O.ai, Navdeep contributed to initiatives spanning machine learning interpretability, automated machine learning (AutoML), and GPU-accelerated model training. He played a key role in embedding interpretability into H2O's flagship AutoML platform, Driverless.ai, and was instrumental in the development of H2O AutoML, H2O4GPU, and H2O-3 — projects that advanced distributed and accelerated machine learning for industry-scale data science.
In earlier work at Cisco and FICO, Navdeep designed and deployed machine learning and analytics solutions across financial services, telecommunications, and market research.
Before entering the tech industry, Navdeep conducted research in cognitive neuroscience and visual psychophysics. At the University of California, San Francisco, he studied neural mechanisms of memory and attention, focusing on how these functions change with aging and dementia. At the Smith-Kettlewell Eye Research Institute, he explored how the brain perceives depth in 3D space — focusing on how brain injuries shape visual perception and eye movement control.
How I work
Operating principles
A few habits that shape how I build products and partner with engineering and research teams.
P / 01
Clarity
Frame the problem, the user, and the bet plainly — before chasing solutions.
P / 02
Rigor
Measure what matters. Decisions are stronger when grounded in data and direct user feedback.
P / 03
Ownership
Clear accountability across product, engineering, and research — from idea to launch and beyond.
P / 04
Resilience
Build for real-world conditions — edge cases, drift, and the messy realities of production.
02 —
-
2024 — Present
ServiceNow
Santa Clara, CA
- Senior Manager, Product Management Jan 2026 — Present
- Staff Senior Product Manager, Responsible AI 2024 — 2026
-
2015 — 2024
H2O.ai
Mountain View, CA
- Manager · Lead Data Scientist2021 — 2024
- Senior Software Engineer2018 — 2021
- Software Engineer2015 — 2018
-
2014 — 2015
Cisco
San Jose, CA
- Data Scientist2014 — 2015
-
2013 — 2014
FICO
San Rafael, CA
- Analytic Science Consultant2013 — 2014
Earlier — Research
Smith-Kettlewell Eye Research Institute
San Francisco, CA
Research Intern
2011
03 —
California State University, East Bay
- M.S. Statistics — Designated Emphasis in Computational Statistics 2014
- B.S. Statistics · B.A. Psychology · Minor, Mathematics 2012
04 —
Current
-
01
Responsible AI at ServiceNow Responsible AI initiatives at ServiceNow.
Past
-
01
Responsible AI at H2O.ai Responsible AI initiatives at H2O.ai.
-
02
AI Governance at H2O.ai AI governance initiatives at H2O.ai.
-
03
Driverless AI H2O.ai's flagship automatic machine learning platform.
-
04
H2O Model Security Evaluate and analyze the security of H2O Driverless AI models.
-
05
H2O-3 Open-source, in-memory, distributed, scalable machine learning platform.
-
06
H2O AutoML H2O.ai's automatic machine learning platform.
-
07
H2O4GPU GPU-accelerated machine learning with APIs in Python and R.
-
08
rsparkling R interface for H2O Sparkling Water — H2O-3 algorithms with Apache Spark.
05 —
Featured · NeurIPS 2022
A Brief Overview of AI Governance for Responsible Machine Learning Systems
Watch talk
Ideas on Machine Learning Interpretability
Driverless AI Hands-On: Machine Learning Interpretability
Secure Machine Learning
Interpretability for Generative AI
Actionable Strategies for Mitigating Risks & Driving Adoption with Responsible ML
Responsible Machine Learning with H2O Driverless AI
Deep Dive into Responsible Machine Learning with H2O Driverless AI
Fairness in AI and Machine Learning
From R Script to Production Using rsparkling
No talks tagged with that topic — try another filter.
06 —
Featured Books
O'Reilly · 2020
Responsible Machine Learning
RML
O'Reilly · 2nd Ed.
An Introduction to Machine Learning Interpretability
MLI
O'Reilly · 1st Ed.
An Introduction to Machine Learning Interpretability
MLI
H2O.ai · Report
Guidelines for Effective AI Governance
GOVAll Publications
-
2023ReportGill, N., Zhang, S. Guidelines for Effective AI Governance with Applications in H2O AI Cloud. H2O.ai.
-
2022WorkshopGill, N., Mathur, A., Conde, M. A Brief Overview of AI Governance in Responsible Machine Learning Systems. NeurIPS Workshop on Trustworthy and Socially Responsible Machine Learning (TSRML).
-
2020BookHall, P., Gill, N., Cox, B. Responsible Machine Learning: Actionable Strategies for Mitigating Risk and Driving Adoption. O'Reilly Media.JournalGill, N., Hall, P., Montgomery, K., Schmidt, N. A Responsible Machine Learning Workflow with Focus on Interpretable Models, Post-hoc Explanation, and Discrimination Testing. Information, 11(3):137.
-
2019BookHall, P., Gill, N. An Introduction to Machine Learning Interpretability — Second Edition. O'Reilly Media.WorkshopHall, P., Gill, N., Schmidt, N. Proposed Guidelines for the Responsible Use of Explainable Machine Learning. NeurIPS Workshop on Robust AI in Financial Services.
-
2018ArticleHall, P., Gill, N., Meng, L. Testing Machine Learning Explanation Techniques. O'Reilly Media.
-
2017BookletHall, P., Gill, N., Kurka, M., Phan, W. Machine Learning Interpretability with H2O Driverless AI. H2O.ai.JournalVoytek, B., Samaha, J., Rolle, C. E., Greenberg, Z., Gill, N., Porat, S. Preparatory Encoding of the Fine Scale of Human Spatial Attention. Journal of Cognitive Neuroscience, 29, 1302–1310.
-
2012JournalTyler, C.W., Elsaid, A.M., Likova, L.T., Gill, N., Nicholas, S.C. Analysis of Human Vergence Dynamics. Journal of Vision, 12(11):21.
No publications tagged with that topic — try another filter.
07 —
Loading latest posts from Substack…
08 —
-
2025Mohammadi, K., Puri, A., Belanger Albarran, G., Bansal, M., Gill, N., Chenard, Y., Subramanian, S., Brunet, M.-E., Stanley, J. Attack What Matters: Integrating Expert Insight and Automation in Threat-Model-Aligned Red Teaming. Now AI Conference, Santa Clara, CA — Nov 5.
-
2024Gill, N., Montgomery, K. Interpretability for Generative AI. H2O GenAI Day, Atlanta, GA — Jan 23.
-
2023Gill, N. Guardrails for LLMs. H2O Open Source GenAI World, San Francisco, CA — Nov 7.
-
2022Gill, N., Mathur, A. Incorporating AI Governance to Increase Adoption in Business Applications. MLOps World 2022, New York, NY — July 14.
-
2021Gill, N., Tanco, M. Security Audits for Machine Learning Attacks. MLOps World 2021 — June 16.Gill, N. Training Understandable, Fair, Trustable and Accurate Predictive Modeling Systems. Duke Machine Learning Day, Durham, NC — Mar 27.
-
2019Gill, N. Human Centered Machine Learning. Artificial Intelligence Conference, San Jose, CA — Sep 11.Gill, N. Interpretable Machine Learning Using rsparkling. Symposium on Data Science and Statistics, Bellevue, WA — May 31.Gill, N. Practical Machine Learning Interpretability Techniques. GPU Technology Conference, San Jose, CA — Mar 21.
-
2018Gill, N. Distributed Machine Learning with H2O. Joint Statistical Meeting, Vancouver — Aug 1.Gill, N. H2O AutoML. Symposium on Data Science and Statistics, Reston, VA — May 16.Hall, P., Gill, N., Chan, M. Practical Techniques for Interpreting Machine Learning Models. 1st ACM Conference on Fairness, Accountability, and Transparency (FAT*), New York — Feb 23–24.
-
2017Gill, N., Hall, P., Chan, M. Driverless AI Hands-On Focused on Machine Learning Interpretability. H2O World, Mountain View, CA — Dec 11.Gill, N. From R Script to Production Using rsparkling. Spark Summit, San Francisco, CA — June 14.
-
2016Gill, N. Scalable Machine Learning in R with H2O. useR! Conference, Stanford, CA — July 11.
-
2013Voytek, B., Porat, S., Chamberlain, J., Balthazor, J., Greenberg, Z., Gill, N., Gazzaley, A. Examining the Efficacy of the iPad and Xbox Kinect for Cognitive Science Research. 2nd Meeting of ESCNS, Los Angeles — Mar 15–17.Greenberg, Z., Gill, N., Porat, S., Samaha, J., Kader, T., Voytek, B., Gazzaley, A. Increased Visual Cortical Noise Decreases Cued Visual Attention Distribution. 20th Cognitive Neuroscience Society Meeting, San Francisco — Apr 13–16.
-
2012Tyler, C.W., Gill, N., Nicholas, S. Hysteresis in Stereoscopic Surface Interpolation: A New Paradigm. 12th Vision Sciences Society Meeting, Naples, FL — May 11–16.Gill, N., Fencsik, D. Effects of Disruptions on Multiple Object Tracking. California Cognitive Science Conference, UC Berkeley — Apr 28.
-
2011Gill, N., Fencsik, D. Effects of Distractions on Recovery Time. Psychology Undergraduate Research Conference, UC Berkeley — May 1.
09 —
-
USPTO
US 11,922,283
Model Interpretation
Methods and systems for generating human-understandable explanations of machine learning model predictions.
-
USPTO
US 11,386,342
Model Interpretation
Techniques for surfacing the local and global behavior of predictive models for stakeholder review.
Published & presented at
O'Reilly Books
NeurIPS Workshops
ACM FAccT
arXiv
MDPI Information
USPTO
Spark Summit
useR! Conference
JSM
MLOps World
Let's talk
Working on AI or ML products? Let's compare notes.
Open to conversations on product, AI, and machine learning — from research collaborations and advisory work to product reviews and speaking engagements.