Olawale (Wale) Salaudeen

Pronouncing my name (roughly): oh-la-wah-ley (Olawale) or wah-ley (Wale).

My full name is Olawale Salaudeen; effectively everyone calls me Wale. Generally speaking, if you are writing my name down in any non-conversational setting, I prefer Olawale. My last name is pronounced roughly sah-lah-oo-deen.

I share my first name, at least colloquially, with the incredible rapper and lyricist Wale, although our full first names are technically different: Olawale vs. Olubowale.

Correct pronunciation is entirely optional, as long as I know you're referring to me. The pronunciation someone uses often reflects how we know each other, where they first encountered my name, or who they learned it from, which I find kind of fun.

Some people nevertheless feel compelled to work very hard to pronounce it “correctly.” I should note that I cannot quite pronounce my first name correctly in my English accent either. I have to switch to my Yoruba accent. Do with that what you will :)

Postdoc, Microsoft Research

Incoming Assistant Professor at the University of Illinois at Urbana-Champaign in Fall 2027

Olawale (Wale) Salaudeen is an incoming Assistant Professor at the University of Illinois at Urbana-Champaign and a Postdoctoral Researcher at Microsoft Research. He was previously an AI Center Fellow at Schmidt Sciences, a Postdoctoral Researcher at MIT in the Healthy ML Lab, and a Postdoctoral Scholar at the Eric and Wendy Schmidt Center at the Broad Institute of MIT and Harvard. He earned a Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign and the Stanford STAIR Lab.

His research is concerned with the scientific foundations of AI systems that have positive societal impact. A central question underlying this goal is one of generalization: when can we know that an AI system will do something we have never directly observed it do in a particular context? He approaches this question by developing the science of valid measurement and prediction of AI capabilities and risks, studying the mechanisms that govern AI behavior and generalization, and developing algorithms and interventions that help AI systems remain reliable as conditions change. His work has appeared in leading AI and machine learning venues and received a Best Paper Award at the NeurIPS 2025 Workshop on Evaluating the Evolving LLM Lifecycle.

He has received a Sloan Scholarship, a Beckman Graduate Research Fellowship, a GEM Associate Fellowship, and an NSF Miniature Brain Machinery Traineeship. He has interned at Sandia National Laboratories, Google Brain, Cruise LLC, and the Max Planck Institute for Intelligent Systems. He received a B.S. in Mechanical Engineering from Texas A&M University.

Olawale (Wale) Salaudeen is an incoming Assistant Professor at the University of Illinois at Urbana-Champaign and a Postdoctoral Researcher at Microsoft Research. He was previously at Schmidt Sciences, MIT, and the Broad Institute. He earned a Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign and the Stanford STAIR Lab.

His research is concerned with the scientific foundations of AI systems that have positive societal impact. A central question underlying this goal is one of generalization: when can we know that an AI system will do something we have never directly observed it do in a particular context? He approaches this question by developing the science of valid measurement and prediction of AI capabilities and risks, studying the mechanisms that govern AI behavior and generalization, and developing algorithms and interventions that help AI systems remain reliable as conditions change.

Olawale (Wale) Salaudeen is an incoming Assistant Professor at the University of Illinois at Urbana-Champaign and a Postdoctoral Researcher at Microsoft Research. He earned a Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign and the Stanford STAIR Lab. His research is concerned with the scientific foundations of AI systems that have positive societal impact. A central question underlying this goal is one of generalization: when can we know that an AI system will do something we have never directly observed it do in a particular context?

I am recruiting PhD students and postdocs for Fall 2027!

Read if prospective:
More details on:

Research

I’m broadly interested in generalization in modern AI: when and why systems generalize, what properties of a system tell us about how it will behave in settings we have not directly tested, how we can develop methods that cause systems to generalize according to the objectives and principles we intend, and how all of this informs consequential decisions about the roles AI systems play in society.

I’m especially excited about:

Algorithms for improved generalization

I'm interested in pre- and post-training methods that push models to generalize according to the underlying objectives, principles, or mechanisms we care about, rather than relying on surface-level cues or exploitable proxies. For example, how can we train systems to infer and pursue the intent behind a reward rather than merely optimize for the easiest way to obtain it? More broadly, I want to understand how training objectives, representations, environments, and interventions shape the kinds of generalization that emerge. I'm especially interested in how and if tools from causal representation learning and mechanism identification, distributionally robust optimization, and information theory can help address these questions.

Relevant work on Characterizing and Intervening on Causal and Spurious Mechanisms of AI Behavior
Relevant work on Statistical Foundations of Inference-Time Adaptation of AI Systems
New paradigms of generalizable AI evaluation

I'm interested in rethinking how we study AI systems, rather than primarily studying the limitations of existing benchmarks or other mechanisms for hill-climbing within current evaluation paradigms. Many of our evaluation norms were developed for very different kinds of systems and use cases, and it is not obvious that their underlying assumptions should continue to hold. This means developing new technical approaches that engage directly with what is distinctive about modern AI systems. For example: evaluation through intervention rather than observation alone, adaptive experiments tailored to each individual system being studied, methods for mapping spaces of possible behavior rather than performance on a fixed task distribution, and approaches that combine observational and interventional evidence.

Relevant work on Valid Measurement and Prediction of AI Capabilities and Risks
AI and society

I'm interested in how our understanding of generalization should inform consequential decisions about AI systems: when we should trust models outside the settings in which they were developed, how uncertainty about generalization should shape deployment and policy decisions, and what technical, institutional, and monitoring infrastructure is needed to mitigate the effects of generalization failures in high-stakes domains such as health and medicine.

Relevant work on Measuring and Steering AI's Societal Impacts

See all publications →

Selected Recent News

Summer 2026. position I will join the University of Illinois at Urbana-Champaign as an Assistant Professor in the Department of Electrical and Computer Engineering, with affiliate appointments in the Coordinated Science Laboratory and the Siebel School of Computing and Data Science, beginning Fall 2027.
Summer 2026. service I am co-organizing the COLM workshop on AI Measurement Science: Toward Rigorous AI Evaluation (led by Sang Truong). Looking forward to seeing your submissions and attendance!
Summer 2026. service I am co-organizing the NeurIPS competition Predictive AI Evaluation Competition (led by Sang Truong). Looking forward to seeing your submissions!
Summer 2026. talk I gave a guest lecture at the Harvard T.H. Chan School of Public Health Program in Clinical Effectiveness.
Summer 2026. position I joined the FATE group at Microsoft Research (NYC) as a postdoctoral researcher.
Summer 2026. talk I gave a lecture at the inaugural AHLI Health AI Summer Camp at the University of Washington.
Older News
Spring 2026. talk I gave guest lectures on the science of measurement and evaluation at Cornell Tech, Princeton, and Stanford.
Spring 2026. paper Two papers accepted to the ICML 2026 Position Paper Track: (i) Stop Evaluating AI with Human Tests, Develop Principled, AI-specific Tests instead and (ii) Position: Benchmarks Cannot Establish Deployment Readiness of Clinical AI.
Fall 2025. paper Our NeurIPS spotlight paper, Aggregation Hides OOD Generalization Failures from Spurious Correlations, was featured in MIT News.
Fall 2025. paper Our work On Evaluating Methods vs. Evaluating Models received a best paper award at the NeurIPS Workshop on Evaluating the Evolving LLM Lifecycle.
Fall 2025. paper Our policy brief on validating claims about AI is now available.
Fall 2025. paper Three papers accepted to NeurIPS 2025 (main track), including one spotlight: (i) Aggregation Hides OOD Generalization Failures from Spurious Correlations (Spotlight), (ii) On Group Sufficiency Under Label Bias, and (iii) Understanding challenges to the interpretation of disaggregated evaluations of algorithmic fairness.
Fall 2025. paper Two papers accepted at the NeurIPS Workshop on Evaluating the Evolving LLM Lifecycle, including one oral: (i) On Evaluating Methods vs. Evaluating Models (Oral) and (ii) Measurement to Meaning: A Validity-Centered Framework for AI Evaluation.
Fall 2025. service I am co-organizing the workshop on The Science of Benchmarking and Evaluating AI at EurIPS 25 in Copenhagen.
Fall 2025. paper Our paper Improving Single-round Active Adaptation: A Prediction Variability Perspective is accepted at TMLR.
Summer 2025. paper Our paper on the limitations of domain generalization benchmarks — Are Domain Generalization Benchmarks with Accuracy on the Line Misspecified? — is accepted at TMLR with J2C certification (ICLR 2026 Journal Track).
Summer 2025. paper Our preprint Stop Evaluating AI with Human Tests, Develop Principled, AI-specific Tests instead is now available on arXiv.
Summer 2025. paper Our preprint Understanding challenges to the interpretation of disaggregated evaluations of algorithmic fairness is now available on arXiv.
Summer 2025. service I am serving as a program chair for the Machine Learning for Health (ML4H) conference in San Diego.
Summer 2025. position I will spend the next year at Schmidt Sciences in NYC as a Visiting AI Scientist and AI Center Fellow in Residence.
Spring 2025. position I joined the Eric and Wendy Schmidt Center at the Broad Institute of MIT and Harvard as a postdoctoral scholar.
Spring 2025. paper Our paper Toward an Evaluation Science for Generative AI Systems appeared in The Bridge (National Academy of Engineering).
Spring 2025. talk I gave a talk on addressing distribution shifts at the MIT LIDS Postdoc NEXUS meeting.
Winter 2025. service I am co-organizing the AI for Society seminar at MIT.
Winter 2025. paper Our paper What's in a Query: Polarity-Aware Distribution-Based Fair Ranking will appear at WWW 2025.
Winter 2025. honors I was selected as an NYU Tandon Faculty First-Look Fellow.
Winter 2025. service I am co-organizing the 30th Annual Sanjoy K. Mitter LIDS Student Conference at MIT.
Winter 2025. honors I was selected as a Georgia Tech FOCUS Fellow.
Fall 2024. paper Our paper On Domain Generalization Datasets as Proxy Benchmarks for Causal Representation Learning will appear at the NeurIPS 2024 Workshop on Causal Representation Learning as an oral presentation.
Fall 2024. position I joined the Healthy ML Lab, led by Prof. Marzyeh Ghassemi, at MIT as a postdoctoral associate.
Spring 2025. paper Our preprint on domain generalization benchmarks — Are Domain Generalization Benchmarks with Accuracy on the Line Misspecified? — is now available on arXiv.
Summer 2024. talk I gave a talk on our work on distribution shift at Texas State's Computer Science seminar.
Summer 2024. talk I gave a talk on our work on distribution shift at UT Austin's Institute for Foundations of Machine Learning (IFML).
Summer 2024. honors I successfully defended my PhD dissertation, "Towards Externally Valid Machine Learning: A Spurious Correlations Perspective."
Spring 2024. talk I gave a talk on AI for critical systems at the MobiliT.AI forum.
Spring 2024. talk I gave a talk at UIUC Machine Learning Seminar on our work on the external validity of ImageNet.
Spring 2024. paper Our preprint ImageNot: A contrast with ImageNet preserves model rankings is now available on arXiv.
Winter 2024. paper Two papers on machine learning under distribution shift will appear at AISTATS 2024.
Fall 2023. position I joined the Social Foundations of Computation department at the Max Planck Institute for Intelligent Systems as a Research Intern working with Dr. Moritz Hardt.
Spring 2023. honors I passed my PhD Preliminary Exam.
Spring 2023. position I will join Cruise LLC's Autonomous Vehicles Behaviors team as a Machine Learning Intern.
Fall 2022. position I moved to Stanford University as a visiting student with Prof. Sanmi Koyejo.
Summer 2022. honors Honored to be selected as a top reviewer (10%) of ICML 2022.
Summer 2022. position Joining Google Brain (now Google DeepMind) in Cambridge, MA as a Research Intern.
Fall 2021. paper Our paper Exploiting Causal Chains for Domain Generalization was accepted at the NeurIPS 2021 Workshop on Distribution Shifts.
Fall 2021. honors Selected as a Miniature Brain Machinery (MBM) NSF Research Trainee.
Summer 2021. honors I was selected to receive an Illinois GEM Associate Fellowship.
Spring 2021. honors I passed my Ph.D. qualifying exam.
Spring 2020. honors I was selected to receive a 2020 Beckman Institute Graduate Fellowship.
No matching items

Experience

2015–19 Texas A&M B.S.
2017–22 Sandia R&D Intern
2019–24 Illinois PhD w/ Sanmi Koyejo
2022 Google Brain Student Researcher w/ Alex D'Amour
2022–24 Stanford PhD w/ Sanmi Koyejo
2023 Cruise ML Intern
2023–24 Max Planck Research Intern w/ Moritz Hardt
2024– MIT Postdoc w/ Marzyeh Ghassemi
2025–26 Broad Institute Postdoc
2025–26 Schmidt Sciences AI Center Fellow
2026– Microsoft Research Postdoc w/ Solon Barocas, Hanna Wallach & Jenn Wortman Vaughan
2027 Illinois Incoming
Faculty

Selected Honors

  • Schmidt Sciences AI Center Fellowship
  • Postdoc Scholar, Eric and Wendy Schmidt Center, Broad Institute
  • Best Paper, NeurIPS 2025 Workshop on Evaluating the Evolving LLM Lifecycle
  • NeurIPS 2025 Top Area Chair
  • NYU Tandon Faculty First-Look Fellow
  • Georgia Tech FOCUS Fellow
  • Sloan Scholarship
  • Beckman Graduate Research Fellowship
  • GEM Associate Fellowship
  • NSF Miniature Brain Machinery Traineeship
  • ICML 2022 Top Reviewer (10%)

Selected Research Mentoring

Current

Previous

  • Lena Stempfle (Visiting Ph.D. Student @ MIT, 2024–25) → Postdoc, MIT
  • Lucia Huo (Master’s Student @ MIT, 2024–25)
  • Omar Dahleh (MEng @ MIT, 2024–25) → Software Engineer, Lumos
  • Sahal Ahmed (UROP @ MIT, 2024–25)
  • Shiny Weng (M.S. @ Stanford, 2024–25)
  • Vivien Jiang (BSRP @ the Broad Institute of MIT and Harvard, summer 2025)
  • Uzma Hamid (LINXS @ Stanford University, summer 2024)
  • Vikram Duvvur (Undergrad @ UIUC, 2021–22) → MS in Machine Learning @ CMU
  • Ahmed Elsayed (DREU @ UIUC, summer 2021) → Software Engineer, Microsoft

Fun Facts

I was born in Nigeria and moved to Dallas, Texas at a young age. I played basketball and water polo in high school. I'm a loyal (if perpetually disappointed) Dallas sports fan, a cinephile, a social dancer (Latin and swing mostly), and a regular at live standup comedy. I'm full of hot takes, most of which I wouldn't die on, but am always eager to share and defend for fun. I also once won an intramural cornhole championship.