Profile pictureHi I am Luis Oala. I work on composable systems for measuring, optimizing and exchanging data states across the entire data generating life-cycle in machine learning.

In regular intervals, I share my ideas through writing, code and presentations spanning topics such as data optimization [1, 2, 3, 4, 5], ML data formats [1, 2, 3] or measurement tools for ML systems [1, 2, 3, 4, 5].

I am a big proponent, maintainer and user of open source machine learning infra. I co-initiated the ML metadata format Croissant now used by most of the big ML data hosters (Huggingface, Kaggle, Dataverse, and others), chair the industry working group Data at the ITU to reduce transaction costs in data systems and support my friends at OpenML as a core member.

I also enjoy promoting opportunities for community across engineering and the arts. I helped initiate machine learning venues such as Data-Centric Machine Learning Research (DMLR) and AI for Good and co-chaired conferences such as ICLR, the DMLR workshop series or ML4H. Together with my brother I run the music label Elektroinstallation Monopohl.

A core thesis behind my work is that a lot of ML systems bottlenecks collapse into fast, granular access problems to the right data. That is also the premise behind Brickroad, where I am building the data multiplexer as Co-Founder and CTO, a technology that lets you access frontier data at the rate of compute. I am also a final-year PhD research scientist at the Department of Artificial Intelligence of Wojciech Samek at Fraunhofer HHI in Berlin, Germany. Previously, I built Dotphoton in Switzerland, data infrastructure for imaging machine learning workloads at petabyte scale.

Writing

See Google Scholar

Talks and Presentations
  • 2026.09.18 | Navigating the Data Frontier with the Data Multiplexer | [program] | Invited Lecture @ National AI Application Base | Hangzhou, China
  • 2026.08.06 | Student Mentorship | [program] | Mentor @ Deep Learning Indaba 2026 | Lagos, Nigeria
  • 2026.07.11 | AI Should Verify, Not Judge, Scientific Work | [program] | Oral @ ICML AI Scientists Workshop | Seoul, South Korea
  • 2026.06.27 | AI Supply Chains | [program] | Panel Talk @ ACM FAccT | Montreal, Canada
  • 2026.05.05 | The Data Multiplexer: Flywheel for Data Flow in the Agent Economy | [program] | Invited Talk @ Datamakers Fest | Porto, Portugal
  • 2026.03.27 | Data standards for health AI: Benchmarking, metadata and federated data discovery | [program] | Invited Talk @ AI for Good | Geneva, Switzerland
  • 2026.02.03 | On Roman Roads: Data Composition For AI & Agents | [program] | Invited Talk @ MIT CSAIL | Boston, USA
  • 2025.12.03 | A Sustainable Machine Learning Economy Needs Data Deals That Work for Generators | [program] | Poster @ NeurIPS 2025 | San Diego, USA
  • 2025.12.01 | The AI Economy: Flywheels for Agentic Value Creation | [program] | Talk @ MLCommons Endpoints 2025 | San Diego, USA
  • 2025.09.18 | A Fever Dream of Machine Learning Framework Composability: Croissant and Beyond | [program] | Invited Talk @ 2025 Berlin Summer School of Artificial Intelligence and Society | Berlin, Germany
  • 2025.08.24 | Intro to Data-Centric Deep Learning | [program] | Invited Lecture @ 2025 TAIK AI Camp | Ethiopia, Cameroon, and Tanzania
  • 2025.03.17 | Data Market for Healthcare AI | [slides] [notes] | Invited Talk @ National University Singapore | Singapore
  • 2024.12.13 | Croissant: A Metadata Format for ML-Ready Datasets | [abstract] [slides] [poster] | Posterspotlight @ NeurIPS 2024 | Vancouver, Canada
  • 2024.12.11 | Generative Fractional Diffusion Models | [abstract] [slides] [poster] | Poster @ NeurIPS 2024 | Vancouver, Canada
  • 2024.12.04 | A Fever Dream of Machine Learning Framework Composability | [abstract] [video] | Invited Talk @ Microsoft Research | Nairobi, Kenya
  • 2024.09.17 | Dotphoton: Your Image Data, Fit for AI | [slides] | Invited Talk @ Innosuisse | San Francisco, USA
  • 2024.09.02 | Paradoxes in Data-Centric Machine Learning | [slides] [abstract] | Invited Talk @ Deep Learning Indaba 2024 | Dakar, Senegal
  • 2024.06.09 | Croissant: A Metadata Format for ML-Ready Datasets | [paper] [slides] | Contributed Talkbest paper award @ SIGMOD/PODS Data Management for End-to-End Machine Learning Workshop | Santiago, Chile
  • 2024.05.31 | From Diverse Datasets to United Nations Public Good Tasks | [program] [slides] | Opening Remarks @ ITU AI for Good Summit | Geneva, Switzerland
  • 2024.01.12 | DTX (Data-Transform Exchange): A Protocol for Composable Data and Transform Transactions | [slides] | Session Chair @ Schloss Dagstuhl Open Machine Learning 2024 Winter Workshop | Dagstuhl, Germany
  • 2023.12.12. | DiffInfinite: Large Mask-Image Synthesis via Parallel Random Patch Diffusion in Histopathology | [abstract] | Posterspotlight @ NeurIPS 2023 | New Orleans, USA
  • 2023.11.30 | Metrological Machine Learning (2ML) | [slides] [program] | Invited Talk @ IEEE BIP Tecnológico de Costa Rica | San Carlos, Costa Rica
  • 2023.11.01 | Metrological Machine Learning (2ML) | [slides] [report] | Invited Talk @ King Abdulaziz City for Science and Technology (KACST) | Riyadh, Saudi Arabia
  • 2023.10.02 | Inspiration Exchange - Data-Centric AI | [recording] | Panelist @ Mihaela van der Schaar Lab University of Cambridge | Cyberspace
  • 2023.07.12 | Interview | [abstract] [raw video] | Interview @ IEEE TEMS/ACM with Stephen Ibaraki | Cyberspace
  • 2023.03.24 | Data and AI solution Assessment Methods | [slides] | Invited Talk @ Harvard University | Cambridge, USA
  • 2023.01.10 | DMLR: Data-centric Machine Learning Research - Past, Present and Future | [notes] | Session Chair @ Asilomar Retreat on Future of Datasets | Monterey, USA
  • 2022.11.28 | Q&A: Emmanuel Candes | [recording] [program] | Session Chair @ ML4H 2022 | New Orleans, USA
  • 2022.11.28 | Q&A: Ben Recht | [recording] [program] | Session Chair @ ML4H 2022 | New Orleans, USA
  • 2022.11.28 | Panel with Himabindu Lakkaraju, Zack Lipton & Mihaela van der Schaar | [recording] [program] | Session Chair @ ML4H 2022 | New Orleans, USA
  • 2022.05.23 | The Audit of a Diabetic Retinopathy Classification Model | [poster] | Poster @ SAIL 2022 | Hamilton, Bermuda
  • 2021.03.17 | Interval Neural Networks as Instability Detectors for Image Reconstructions | [recording] [program] [paper] | Contributed Talkbest paper award @ BVM 2021 | Regensburg, Germany
  • 2020.12.11 | ML4H Auditing: From Paper to Practice | [recording] [poster] [paper] | Contributed Talkspotlight @ ML4H 2020 | Cyberspace
  • 2020.07.17 | Detecting Failure Modes in Image Reconstructions with Interval Neural Network Uncertainty | [recording] [paper] [program] | Contributed Talkspotlight @ ICML UDL 2020 | Cyberspace
  • 2020.01.22 | AI Test Metric Specification | [slides] | Invited Talk @ WHO PAHO | Brasilia, Brazil
  • 2019.11.12 | Data and AI Solution Assessment Methods | [slides] | Invited Talk @ Department of Telecommunications of India | New Delhi, India
  • 2019.09.04 | Data and AI Solution Assessment Methods | [slides] | Invited Talk @ Universal Communications Service Access Fund of Tanzania | Zanzibar, Tanzania