About Me

I work on LLMs (OLMo and olmOCR) at Ai2. My days are spent implementing the latest research findings, building datasets and benchmarks, and convincing neural networks to behave. In short, I help build language models that (usually) do what they're told with minimal grumbling.

When I am bored, I implement latest research findings into LLMs for fun.

I'm all about adventure whether it's exploring new places, experimenting in the kitchen, playing sports, writing, or getting lost in movies.

Research Interests

Pioneering the fusion of NLP and Generative AI, I strive to solve complex challenges. My research zeroes in on three crucial areas:
1. Attention Mechanisms.
2. Training LLMs.
3. Dataset Quality, Bias and Fact-Checking.

Experience

Allen Institute for AI (Ai2)
LLM Solution Architect, October 2024 - Present.

Data Mining Lab, Illinois Institute of Technology
Machine Learning Research Associate, August 2024 - October 2024.
Research Assistant, December 2022 - May 2024.

Research in collaboration with: IARPA, Cisco, U.S. National Science Foundation, CAOE.

Appyhub Technology Solutions
Image Processing Engineer, October 2020 - April 2022.

Artificial Intelligence and Robotics Center, VIT-AP
Research Assistant, August 2018 - May 2020.

Recent Research

olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models
Jake Poznanski, Aman Rangapur, Kyle Lo, Luca Soldaini
ICML CODEML Workshop 2025
[Technical Paper], [GitHub], [Hugging Face], [Demo]

2 OLMo 2 Furious
Team OLMo
COLM 2025
[Technical Paper], [GitHub], [Hugging Face], [Blog]

Piecing It All Together: Verifying Multi-Hop Multimodal Claims
Haoran Wang, Aman Rangapur, Xiongxiao Xu, Yueqing Liang, Kai Shu.
COLING 2025
[Paper], [Leaderboard], [GitHub]

Fin-Fact: A Benchmark Dataset for Multimodal Financial Fact Checking and Explanation Generation
Aman Rangapur, Haoran Wang, Kai Shu.
ACM Web Conference 2025
Fin-Fact was used at FMD workshop at COLING 2025.
[Paper], [Presentation], [GitHub], [Hugging Face], [Leaderboard]

Investigating Online Financial Misinformation and Its Consequences: A Computational Perspective
Aman Rangapur, Haoran Wang, Kai Shu.
ACM Computing Surveys(Submission Under Review)
[Pre Print]

EfficientWord-Net: An Open Source Hotword Detection Engine Based on Few-Shot Learning
Chidhambararajan R, Aman Rangapur, S. Sibi Chakkaravarthy.
Journal of Information & Knowledge Management
[Paper], [GitHub]

Academic Service

Program Commitee MisD 25.
Sub-reviewer of AAAI 2024 conference.
Sub-reviewer of NeurIPS 2023 conference.
Reviewer of International Journal of Computing and Digital Systems 2023 conference.
Volunteer for ACM FAccT Conference 2023.

Fellowships & Awards


Graduate Student Scholarship	Illinois Institute of Technology
Admission Merit Scholarship	George Mason University
Runner-up position	Major League Hacking 2018