Huzefa Rangwala, PhD

Staff Scientist, Physical AI, Siemens

About

Science leader with a unique blend of academic rigor and industry impact. Previously, I led an applied science organization at Amazon Web Services (AWS), driving post-training science for Amazon Bedrock and SageMaker — from customizing foundation models and developing guardrails, to leading open-source efforts on graph ML and AutoML. Prior to that, I held a tenured professorship in the Computer Science Department at George Mason University (2008–2025), including a stint as Interim Chair (2021–2022). I am most proud to be the recipient of the NSF CAREER Award (2013–2018), the GMU Teaching Excellence Award (2014), the Outstanding Researcher Award (2018), and the OSCAR Undergraduate Mentoring Excellence Award (2018). Thanks to NSF, NIH, DARPA, Google, DHS, and NRL for supporting my research. I hold a Ph.D. (2008) and M.S. (2005) from the University of Minnesota, Twin Cities and a B.E. from VJTI Mumbai (2003).

What energizes me most is connecting advanced science to real problems. In academia, this meant pursuing inter-disciplinary science — applying machine learning to biology, medicine, education, and social sciences — in the belief that the most interesting questions live at the boundaries between fields. In industry, it means connecting scientific innovation to real customer impact. Whether it's training foundation models for time series forecasting, making generative AI safer through guardrails, or optimizing GPU kernels with reinforcement learning — I believe the best research happens when curiosity meets customer obsession.

I'm passionate about building high-performing teams, mentoring the next generation of scientists, teaching, and creating cultures where exploration and delivery coexist. See my student advisees.

News

Launch — Amazon Bedrock reinforcement fine-tuning adds support for open-weight models with OpenAI-compatible APIs. Expanding model customization to Qwen and OpenAI GPT-OSS variants via rule-based and AI-powered reward functions.

ICLR 2026 — Relatron: Automating Relational Machine Learning over Relational Databases. Congratulations to Zhikai Chen for driving this work end-to-end. Co-authors: Han Xie, Haiyang Yu, Xiang Song, and Huzefa Rangwala.

ICLR 2026 — VeriCoT: Neuro-symbolic Chain-of-Thought Validation via Logical Consistency Checks. Congratulations to Yu (Annie) Feng (AWS summer intern) and co-authors Nathaniel Weir, Kaj Bostrom, Sam Bayless, Darion Cassel, Sapana Chaudhary, Benjamin Kiesl-Reiter, and Huzefa Rangwala.

Release — GraphStorm 0.5.1 (December 2025). Adds edge feature support across all built-in GNN models (GraphSAGE, GAT, GATv2), real-time inference with raw text input, and experimental Mitra integration for tabular feature embeddings.

Release — AutoGluon 1.5.0 (December 2025). Introduces Chronos-2 for zero-shot time series forecasting, new tabular models (RealTabPFN-2, TabDPT), and up to 80% win rate over v1.4 on time series benchmarks.

Launch — Chronos-2: From Univariate to Universal Forecasting (October 2025). Amazon's next-generation time series foundation model supporting multivariate, covariate-informed, and zero-shot forecasting. Achieves 90%+ win rate over its predecessor; available open-source on Hugging Face and Amazon SageMaker.

Prior Work Experience

Amazon Web Services — Senior Manager, Applied Science (2022–2026)

Led a 70-person applied science organization driving post-training science for Amazon Bedrock and SageMaker, and open-source efforts in GraphML and AutoML.

2025–26
My teams shipped several high-impact launches. We released Chronos-2, Amazon's foundation model for time series forecasting, which achieved state-of-the-art results across all major benchmarks and surpassed 600M downloads. AutoGluon 1.4 and 1.5 set new records for tabular and time series prediction. I led science for Amazon Bedrock Guardrails, expanding content moderation to 100+ languages and new modalities. We launched reinforcement fine-tuning at re:Invent 2025, and explored RL-based GPU kernel optimization achieving up to ~2x speedups.

2024
The team launched Amazon Bedrock GraphRAG at re:Invent 2024 and I stood up the Think Forward Lab for forward-looking research in autonomous agents and kernel optimization.

2023
I supported the open-source GraphStorm project for training large-scale graph neural networks, integrated with Amazon Neptune & applied on applications at Amazon for Search, Ads and Fraud.

I led alignment science (RLHF) for Amazon Titan, owned science for Amazon DataZone's data cataloging and governance features, and trained biological sequence models for drug discovery.

2022
I led the team that built custom ML solutions with AWS customers in healthcare, semiconductor, and pharmaceutical industries through the ML Solutions Lab.

George Mason University — Professor of Computer Science (2008–2024)

Interim Chair (2019–2020)
Recruited 13 new faculty members including chair and 10 tenured/tenure-track faculty. Improved CS Rankings score from 60+ to 34. Won a $3M Break Through Tech Grant from Cornell Tech to improve women graduates in CS from 18% to 33%. Led virtual transition during COVID-19 serving 2,000+ students.

Lawrence Cranberry Faculty Fellow (2019–2024)
Endowed faculty fellowship recognizing research excellence. Also served as Smithsonian Institution Fellow (2021).

Research & Funding
Secured over $10M in external funding as PI/Co-PI from NSF, NIH, DARPA, Google, DHS, and NRL. Led research in machine learning, data mining, bioinformatics, learning analytics, and fairness-aware algorithms. Published 130+ peer-reviewed papers. General Chair for ACM KDD 2022 in Washington, DC (2,000+ attendees).

Teaching & Mentoring
Graduated 12 Ph.D. students (now at Google, Microsoft, Amazon, Intel, LinkedIn). Advised 7 M.S. thesis students and 14 undergraduate mentees. Taught Data Mining, Computer Systems Architecture, Biological Data Mining, and Parallel Computing. Received Outstanding Researcher, Outstanding Teaching, and Undergraduate Mentoring Excellence awards. NSF CAREER Award recipient (2013–2018).

Student Advisees

Ph.D. Alumni

† Winner of CS Department Outstanding Graduate Student Award

Name	Thesis	Graduated	Position After
Jonathan Vasquez	Auditing Discrimination Risks in Datasets	12/2024	Co-Founder, EvoConsulting & EvoAcademy
Xavier Gitiaux	Pre- and Post-Fairness Processing for Black Box Classifiers	05/2022	Scientist, Microsoft
Al-Amin Hosain (Co-Advisor: Prof. Kosecka)	Word-Level Sign Language Recognition From Videos	12/2021	Scientist, Walmart Labs
Yujing Chen	Sensor Analytics using Multi-Task Learning and Federated Learning	05/2021	Applied Scientist, Amazon
Mohammad Arifur Rahman	Efficient and Interpretable Machine Learning Algorithms for Predictive Analyses in Metagenomic Data	12/2020	Applied Scientist, Amazon
Qian Hu	Educational Data Mining for All: Fair Machine Learning for Student Modeling and Forecasting	07/2020	Applied Scientist, Amazon
Zhiyun Ren	Academic Performance Prediction Using Machine Learning Techniques	05/2019	Scientist, LinkedIn
Azad Naik	Hierarchical Classification with Rare Categories and Inconsistencies	02/2017	Data Scientist, Microsoft
Anveshi Charuvaka †	Regularized Learning in Multiple Tasks With Relationships	09/2017	Applied Scientist, Amazon
Tanwistha Saha (Co-Advisor: Prof. Domeniconi)	Learning in Relational Networks	11/2014	Advanced Analytics & Deployment Engineer, Intel
Samuel Blasiak †	Latent Variable Models of Sequences for Classification and Discovery	08/2013	Software Engineer, Google
Zeehasham Rasheed	Data Mining Framework for Metagenome Analysis	04/2013	Data Scientist, AOL
Nuttachat Wisittipanit (Co-Advisor: Prof. Gillevet)	Machine Learning Approach for Profiling Human Microbiome	03/2012	—

M.S. Thesis Advisees

Mackenzie Sweeney — M.S. in CS (Spring 2015–Summer 2016). Thesis: Intelligent Degree Planning Systems: Advancements in Personalized Achievability Rating.
Nikhil Muralidhar — M.S. in CS (Spring 2014–Fall 2015). Thesis: Recommending Temporally Relevant News Content Using Implicit Feedback Data: A Tag-based Approach.
Azad Naik — M.S. in CS (Fall 2012–Spring 2013). Thesis: Using Multi-Task Learning for Large-Scale Document Classification.
Syed Mahmood — M.S. in CS (Fall 2009–Fall 2012). Thesis: GPU-based Eulerian Assembly of Genomes.
Ammar Naqvi — M.S. in BINF (Co-Advised with Prof. Gillevet, Spring 2009–Fall 2009). Thesis: Understanding and Analyzing the Human Microbiome.
Salman Jamali — M.S. in CS (Spring 2009–Summer 2009). Thesis: Comment Mining, Popularity Prediction and Social Network Analysis.

Undergraduate Mentees

Nathan LaPierre, B.S. in CS (Fall 2015–Summer 2016) — Sponsor: NSF
Abigail Justen, B.S. in CS (Summer 2016) — Sponsor: NSF BIGDATA
Thi Nguyen, B.S. in CS (Summer 2016) — Sponsor: NSF BIGDATA
Ameer Takaddein, B.S. in CS (Summer 2016) — Sponsor: NSF BIGDATA
Jason Ko, B.S. in CS (Summer 2016) — Sponsor: NSF BIGDATA
Michel Jean Rouly, B.S. in CS (Summer 2015) — Sponsor: NSF
Anisha Kolla, B.S. in CS (Spring 2013) — Sponsor: URSP, OSCAR, GMU
Hozaifah Zafar, B.S. in CS (Spring 2013–) — Sponsor: USRP, OSCAR, GMU
Emily Eastlake, B.S. in BENG (Summer 2012) — Sponsor: SURE Program, Bioengineering, GMU
Liban Hassan, B.S. in ECE (Summer 2011) — Sponsor: NSF LSAMP Program, GMU
Amanda Zouzolou, B.S. in BENG (Summer 2011) — Sponsor: SURE Program, Bioengineering, GMU
Roderick Tolbert, B.S. in CS (Summer 2011) — Sponsor: NSF REU
Charles Sweet, B.S. in CS (Summer 2010–Fall 2012) — Sponsor: UAP Program, GMU
Minh Bui, B.S. in CS (Summer 2010) — Sponsor: UAP Program, GMU