profile photo


Email  |  Google Scholar  |  Github
LinkedIn  |  Twitter

Akshara Prabhakar


Hi! I am an Applied Scientist at Salesforce AI Research. I received my MS in Computer Science from Princeton University, advised by Dr. Karthik Narasimhan. I was part of the Princeton NLP group and have collaborated with the PLI and CoCoSci labs. Prior to grad school, I graduated as a Gold Medalist from National Institute of Technology Karnataka, Surathkal, advised by Dr. Sowmya Kamath.

I enjoy applied/product research to solve real-world tasks. My current works revolve around benchmarking large language models, language agents and understanding their cognitive aspects.


    I am fortunate to have studied and worked at some wonderful places!

2024 - present
summer 2023
2022 - 2024
summer 2022
summer 2021
2021
summer 2021
2020 - 2022
summer 2020
summer 2020
2019 - 2020
2018 - 2022
2007 - 2018

Sep '24  

 Our paper on understanding the factors influencing CoT is accepted at EMNLP Findings 2024!

Jun '24  

 Joined Salesforce Research as an Applied Scientist.

Sep '23  

 Named a Siebel Scholar (Class of 2024)!

Jun '23  

 Joining Microsoft as an Applied Scientist intern with the Ads team.

Jan '23  

 Invited for Citadel Security Women's Datathon 2023 final onsite held in Miami, FL.

Oct '22  

 Awarded Institute Gold medal by National Institute of Technology Karnataka for securing highest GPA in my batch.

Apr '22  

 Paper accepted in NAACL 2022 main conference!

Feb '22  

 Awarded AAAI 2022 Student Scholarship.

Jan '22  

 My Bachelors thesis work is accepted in CISS 2022!

LoRA Soups: Merging LoRAs for Practical Skill Composition Tasks
International Conference on Computational Linguistics (COLING 2025) - Industry
[Paper] [Code]
Akshara Prabhakar, Yuanzhi Li, Karthik Narasimhan, Sham Kakade, Eran Malach, Samy Jelassi
CRMArena: Understanding the Capacity of LLM Agents to Perform Professional CRM Tasks in Realistic Environments
[Preprint] [Code]
Kung-Hsiang Huang, Akshara Prabhakar, Sidharth Dhawan, Yixin Mao, Huan Wang, Silvio Savarese, Caiming Xiong, Philippe Laban, Chien-Sheng Wu
Deciphering the Factors Influencing the Efficacy of Chain-of-Thought: Probability, Memorization, and Noisy Reasoning
Empirical Methods in Natural Language Processing (EMNLP 2024 Findings); also in BlackboxNLP 2024
[Paper] [Code] [Poster]
Akshara Prabhakar, Thomas L. Griffiths, R. Thomas McCoy
xLAM: A Family of Large Action Models to Empower AI Agent Systems
[Technical Report]
Jianguo Zhang*, Tian Lan*, Ming Zhu*, Zuxin Liu*, Thai Hoang*, Shirley Kokane, Weiran Yao, Juntao Tan, Akshara Prabhakar, Haolin Chen, Zhiwei Liu, Yihao Feng, Tulika Awalgaonkar, Rithesh Murthy, Eric Hu, Zeyuan Chen, Ran Xu, Juan Carlos Niebles, Shelby Heinecke, Huan Wang, Silvio Savarese, Caiming Xiong
Exploring Reasoning and Interactive Benchmarking of Language Models
M.S.E. Thesis, Princeton Univeristy (2024)
[Thesis] [Slides]
Akshara Prabhakar
Language Models as Science Tutors
International Conference on Machine Learning (ICML 2024); also in Data-Centric Machine Learning Research Workshop @ ICLR 2024
[Paper] [Code]
Alexis Chevalier, Jiayi Geng, Alexander Wettig, Howard Chen, Sebastian Mizera, Simon Machado, Arturo Rodriguez Fanlo, Simon Frieder, Zirui Wang, Akshara Prabhakar, Jiachen T. Wang, Xindi Wu, Mengzhou Xia, Wenhan Xia, Jiatong Yu, Ellie Thieu, Max Aragon, Zhiyong Ren, Junjie Zhu, Toni Annala, Sanjeev Arora, Danqi Chen
InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedback
Conference on Neural Information Processing Systems (NeurIPS 2023) - Datasets and Benchmarks
[Paper] [Code] [Poster]
John Yang, Akshara Prabhakar, Karthik Narasimhan, Shunyu Yao
Language Agents as Hackers: Evaluating Cybersecurity Skills with Capture the Flag
Multi-Agent Security Workshop @ Neural Information Processing Systems (NeurIPS 2023)
[Paper] [Code] [Poster]
John Yang, Akshara Prabhakar, Shunyu Yao, Kexin Pei, Karthik Narasimhan
Best Paper Award
Commonsense and Named Entity Aware Knowledge Grounded Dialogue Generation
North American Chapter of the Association for Computational Linguistics (NAACL-HLT 2022)
[Paper] [Code] [Slides]
Deeksha Varshney, Akshara Prabhakar, Asif Ekbal
Neural Language Modeling of Unstructured Clinical Notes for Automated Patient Phenotyping
56th Annual Conference on Information Sciences and Systems (CISS 2022)
[Paper] [Slides]
Akshara Prabhakar, Shidharth S, Sowmya Kamath S
CL-NERIL: A Cross-Lingual Model for NER in Indian Languages
36th AAAI Conference on Artificial Intelligence (AAAI 2022) - Student Poster
[Paper] [Code] [Poster]
Akshara Prabhakar, Gouri Sankar Majumder, Ashish Anand
Student Scholarship Award
Diagnostic Code Group Prediction by Integrating Structured & Unstructured Clinical Data
9th International Conference on Big Data Analytics (BDA 2021)
[Paper] [Video] [Slides]
A short version was accepted at the Young Researchers' Symposium of ACM India KDD CoDS-COMAD 2021!
[Paper] [Slides]
Akshara Prabhakar*, Shidharth S*, Gokul S Krishnan and Sowmya Kamath S
Study of a Hybrid Approach Towards Malware Detection in Executable Files
6th International Conference on Cyber Security and Privacy in Communication Networks 2020
Published in Springer Nature Computer Science journal Volume 2
[Paper]
Akshara Prabhakar, Bhawana Rudra

Teaching Assistant  

 Co-head TA, COS 324: Intro to Machine Learning @ Princeton (Spring '24, Fall '23, Spring '23, Fall '22)

GHC Open-Source Mentor  

 Mentor at Ceph organization
⇧