Abdul Fatir Ansari

Senior Applied Scientist at AWS

prof_pic.webp

Hallo! I am Fatir, a Machine Learning Scientist at Amazon Web Services (AWS) based in Berlin, where I work on time series forecasting, log analytics, and generative models. Of late, I have been working on building general purpose models (aka “foundation” models) for time series problems. I also contribute to AWS’ open source libraries: AutoGluon and GluonTS.

Research: My general research interest lies in the areas of time series analysis and generative models, encompassing probabilistic generative modeling, unsupervised learning, and representation learning. For details about my background and research, please take a look at my curriculum vitae and my google scholar page.

Previously: I graduated with a PhD in Computer Science from NUS School of Computing where I was advised by Prof. Harold Soh and received the Dean’s Graduate Research Excellence Award for my PhD research. Prior to that, I obtained my bachelor’s degree in Civil Engineering from IIT Roorkee. During my undergrad years, I also participated in Google Summer of Code in 2016 and 2017.

Contact: abdulfatirs [at] gmail [dot] com

AWS AI
2021, 2022 - Present
NUS Logo
NUS School of Computing
2017 - 2022
IIT Roorkee
2013 - 2017
Google Summer of Code
S2016, S2017

news

Aug 2025 Delivered a lecture on Building Foundation Models for Time Series at the Oxford ML School.
Jul 2025 Chronos models have crossed 500 million downloads (all time) on Hugging Face!
Jul 2025 Gave a talk on AutoGluon-TimeSeries and GraphStorm at the ICML Expo.
May 2025 Our paper on wavelet-based tokenization for time series foundation mdoels got accepted at ICML.
Jan 2025 Our paper on lightweight covariate adapters for pretrained models got accepted at AISTATS.

latest posts

selected publications

2024

  1. TMLR
    Chronos: Learning the Language of Time Series
    Abdul Fatir AnsariLorenzo StellaCaner TurkmenXiyuan Zhang, Pedro Mercado, Huibin ShenOleksandr Shchur, Syama Syndar Rangapuram, Sebastian Pineda Arango, Shubham Kapoor, Jasper Zschiegner, Danielle C. MaddixHao WangMichael W. Mahoney, Kari Torkkola, Andrew Gordon Wilson, Michael Bohlke-Schneider, and Bernie Wang
    Transactions on Machine Learning Research, 2024
    3.5K Github stars and 500M+ Hugging Face model downloads as of Aug 2025

2023

  1. ICML Oral
    Neural Continuous-Discrete State Space Models for Irregularly-Sampled Time Series
    Abdul Fatir AnsariAlvin Heng, Andre Lim, and Harold Soh
    In International Conference on Machine Learning, 2023
    Oral Presentation (top 2.4%)
  2. NeurIPS
    Predict, Refine, Synthesize: Self-Guiding Diffusion Models for Probabilistic Time Series Forecasting
    Marcel Kollovieh*Abdul Fatir Ansari*, Michael Bohlke-Schneider, Jasper Zschiegner, Hao Wang, and Yuyang Wang
    In Neural Information Processing Systems, 2023

2021

  1. ICLR
    Refining Deep Generative Models via Discriminator Gradient Flow
    Abdul Fatir AnsariMing Liang Ang, and Harold Soh
    In International Conference on Learning Representations, 2021