AI People Wiki

Builders & open source

Rajat Sen

Research Scientist, Google

Rajat Sen is a Research Scientist at Google Research and describes being one of the creators of TimesFM, a pretrained foundation model for time-series forecasting. Sen co-authored the 2023 paper introducing the decoder-only model behind TimesFM, co-wrote the Google Research posts announcing TimesFM in 2024 and TimesFM-3 in 2026, and also works on making large language model inference faster.1234

At a glance

Current role
Research Scientist
Company
Google
Sector
Company
Education
UT Austin
Work listed
3
Sources cited
5
Profile reviewed

Career

Sen obtained a PhD from UT Austin in 2019, advised by Sanjay Shakkottai and working closely with Alex Dimakis, with research on contextual bandits, extreme classification and black-box optimization. Sen is now a Research Scientist at Google Research.1

Research

Sen's current interests are pretrained models for time-series forecasting and other less explored modalities, along with faster inference for large language models. Sen co-authored the TiDE paper (2023) and the decoder-only foundation model paper behind TimesFM (2023), and co-wrote the February 2024 Google Research post that introduced TimesFM.1523

Sen's site records the release of TimesFM-2.0 at the end of 2024. In August 2026, Sen and Ayush Jain introduced TimesFM-3 on the Google Research blog as a zero-shot model for multivariate forecasting, available on GitHub and Hugging Face and integrated into BigQuery.14

Work and contributions

  • TimesFM Open weights

    One of the creators of TimesFM and a co-author of the 2023 paper introducing its decoder-only time-series foundation model; Sen co-wrote the February 2024 Google Research post announcing it.123

  • TimesFM 3.0 Open weights

    Co-author, with Ayush Jain, of the August 2026 Google Research post introducing TimesFM-3, a zero-shot foundation model for multivariate forecasting released on GitHub and Hugging Face.4

  • Long-term Forecasting with TiDE: Time-series Dense Encoder Published research

    Co-author of the 2023 TiDE paper on an encoder-decoder model for long-term time-series forecasting.5

Timeline

  1. Obtained a PhD from UT Austin, advised by Sanjay Shakkottai.1
  2. Co-authored the TiDE paper on long-term time-series forecasting.5
  3. Co-authored the decoder-only time-series foundation model paper behind TimesFM.2
  4. Co-wrote the Google Research post introducing the TimesFM forecasting model.3
  5. Noted on Sen's site that the team had released TimesFM-2.0.1
  6. Co-wrote the Google Research post introducing TimesFM-3 for multivariate forecasting.4

Where they work

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

5 sources · profile reviewed Oct 22, 2026 · data updated Oct 22, 2026

  1. Homepage of Rajat SenRajat Sen · Official page · accessed Oct 22, 2026
  2. A decoder-only foundation model for time-series forecasting (arXiv 2310.10688)arXiv · Paper · published Oct 14, 2023 · accessed Oct 22, 2026
  3. A decoder-only foundation model for time-series forecastingGoogle Research · Announcement · published Feb 2, 2024 · accessed Oct 22, 2026
  4. TimesFM-3: A zero-shot foundation model for multivariate forecastingGoogle Research · Announcement · published Aug 31, 2026 · accessed Oct 22, 2026
  5. Long-term Forecasting with TiDE: Time-series Dense Encoder (arXiv 2304.08424)arXiv · Paper · published Apr 17, 2023 · accessed Oct 22, 2026