Unmasking Text at NeurIPS 2026
Erik Gaasedelen
I lead engineering teams and build machine learning systems.
For over a decade I've worked where research becomes production software. I did a PhD applying deep learning to cardiac anatomy, built computer vision for self-driving cars at Lyft Level 5, and went from senior engineer to Senior Engineering Manager at Woven by Toyota. Now I do AI research and development at Answer.AI.
Where I've worked
- Answer.AI
- Woven by Toyota
- Lyft Level 5
- University of Minnesota

Unmasking Text
An Interactive Guide to Diffusion Language Models
Autoregressive models write one token at a time, left to right. Masked diffusion models start with every position masked and fill in several tokens per step, most confident first. Replay real generation runs from LLaDA-8B-Instruct and Qwen2.5-7B-Instruct, one unmasking step at a time.
- NeurIPS 2026 Education Track
- Poster session Monday 7 December, Sydney
- 9 interactive widgets
- Recorded on one RTX 3090
What I do
-
Research and teach machine learning
My interactive tutorial on diffusion language models, Unmasking Text, is in the NeurIPS 2026 Education Track. At Answer.AI I research and build AI systems, from tools that let coding agents work alongside people to products used by founders and investors.
-
Lead engineering teams
I've set technical direction as a staff engineer and led teams as a Senior Engineering Manager at Woven by Toyota. I still write code.
-
Computer vision and autonomy
At Lyft Level 5 I built deep learning models for computer vision on the mapping team, then spent two years as a software engineer in the self-driving division.
-
Science into software
My PhD at the University of Minnesota's Visible Heart Lab applied deep learning and virtual reality to cardiac anatomy and medical devices. I build things where science and software meet.
Experience
-
Apr 2025 – present
Answer.AI
- R&D Staff
An AI research and development lab. I work on tools for programming with AI, and on AI products.
-
Jul 2021 – Apr 2025
Woven by Toyota
- Senior Engineering ManagerFeb 2024 – Apr 2025
- Staff Software EngineerMay 2023 – Mar 2024
- Senior Software EngineerJul 2021 – Jun 2023
Toyota's software and mobility company.
-
Aug 2018 – Jul 2021
Lyft Level 5
- Software Engineer, autonomous vehiclesMay 2019 – Jul 2021
- Deep learning for computer vision, Mapping Intelligence teamAug – Dec 2018
Lyft's self-driving division.
-
Aug 2015 – Jun 2019
Visible Heart Lab, University of Minnesota
- PhD researcher
Research on cardiac anatomy and medical devices, published in Heart Rhythm and at the Design of Medical Devices Conference.
Research
Papers and tutorials
- Unmasking Text: An Interactive Guide to Diffusion Language ModelsNeurIPS 2026 Education Track
- Algorithm for the analysis of pre-extraction computed tomographic images to evaluate implanted lead–lead interactions and lead–vascular attachmentsHeart Rhythm, 2020
- Virtual reality and visualization of 3D reconstructed medical imaging: learning variations within detailed human anatomiesAdvances in Intelligent Systems and Computing, 2020
- 3D graphics to virtual reality in medicineEngineering in Medicine, 2019
- A deep learning approach for the automatic identification of the left atrium within CT scansDesign of Medical Devices Conference, 2019
- The application of deep learning for the classification of internal human cardiac anatomyDesign of Medical Devices Conference, 2018
- Using smartphone-based virtual reality to explore internal anatomy of 3D heart modelsDesign of Medical Devices Conference, 2017
- Medtronic Reveal LINQ devices provide better understanding of hibernation physiology in the American black bearDesign of Medical Devices Conference, 2017
Education
- PhD, Biomedical Informatics and Computational Biology, University of Minnesota, 2019
- BS, Neuroscience, University of Minnesota, 2014
Also
- Licensed private pilot