thomas

Curriculum Vitae

2021-2022: Research Engineer III
- Smarsh Inc.
- During this time I worked on a speech diarisation system, helped train the speech transcription models and worked on serving these models to customers in a secure way.
2019-2021: Data Scientist (Contract)
- Warner Media

Massively Parallel Expectation Maximization For Approximate Posteriors - Thomas Heap, Sam Bowyer, Laurence Aitchison, , 2025
Sparse Autoencoders Can Interpret Randomly Initialized Transformers - Thomas Heap, Tim Lawson, Lucy Farnik, Laurence Aitchison, arXiv.org, 2025
Using Autodiff to Estimate Posterior Moments, Marginals and Samples - Sam Bowyer, Thomas Heap, Laurence Aitchison, Conference on Uncertainty in Artificial Intelligence, 2023
Massively Parallel Reweighted Wake-Sleep - Thomas Heap, Gavin Leech, L. Aitchison, Conference on Uncertainty in Artificial Intelligence, 2023