Soham Dan

Soham Dan

Bio

I build reliable language and agentic systems with an emphasis on reasoning, robustness, memory, and efficient learning.

I am a Machine Learning Researcher at Scale AI working on LLM agents. Previously, I was a Senior Research Scientist at Microsoft working on Copilot and a Research Scientist at IBM Research. I completed my Ph.D. in Computer and Information Science at the University of Pennsylvania, where I was advised by Prof. Dan Roth in the Cognitive Computation Group. Before that, I earned my undergraduate degree in Computer Science from IIT Kharagpur, where I was awarded the President of India Gold Medal.

Publications

2026

  1. EACL
    Can LLMs reason over extended multilingual contexts? Towards long-context evaluation beyond retrieval over haystacks
    EACL, 2026

2025

  1. NAACL
    Multilingual Needle in a Haystack: Investigating Long-Context Behavior of Multilingual Large Language Models
    NAACL, 2025
  2. ICLR
    Large Language Models can be Strong Self-Detoxifiers
    ICLR, 2025
  3. NAACL
    CTBench: A Comprehensive Benchmark for Evaluating Language Model Capabilities in Clinical Trial Design
    Findings of NAACL, 2025
  4. NAACL
    AssertionBench: A Benchmark to Evaluate Large-Language Models for Assertion Generation
    Findings of NAACL, 2025
  5. COLING
    On the Effects of Fine-tuning Language Models for Text-Based Reinforcement Learning
    COLING, 2025

2024

  1. ICML
    Larimar: Large Language Models with Episodic Memory Control
    ICML, 2024
  2. ICLR
    On the Generalization Capacity of Neural Networks During Generic Multimodal Reasoning
    ICLR, 2024
  3. NAACL
    Language Guided Exploration for RL Agents in Text Environments
    Findings of NAACL, 2024
  4. ACL
    NeuroPrune: A Neuro-inspired Topological Sparse Training Algorithm for Large Language Models
    Findings of ACL, 2024
  5. ACL
    API-BLEND: A Comprehensive Corpora for Training and Benchmarking API LLMs
    ACL, 2024
  6. AAAI
    Generalized Planning in PDDL Domains with Pretrained Large Language Models
    AAAI, 2024

2023

  1. ACL
    MISMATCH: Fine-grained Evaluation of Machine-generated Text with Mismatch Error Types
    Findings of ACL, 2023
  2. AISTATS
    Two-Sample Tests for Inhomogeneous Random Graphs in 𝐿𝑟 norm: Optimality and Asymptotics
    AISTATS, 2023
  3. AISTATS
    One Arrow, Two Kills: An Unified Framework for Achieving Optimal Regret Guarantees in Sleeping Bandits
    AISTATS, 2023
  4. ACL
    In and Out-of-Domain Text Adversarial Robustness via Label Smoothing
    ACL, 2023

2022

  1. ICML
    Understanding Robust Generalization in Learning Regular Languages
    ICML, 2022
  2. CVPR
    Cross-modal Map Learning for Vision and Language Navigation
    CVPR, 2022

2021

  1. NAACL
    Generalization in Instruction Following Systems
    NAACL, 2021
  2. ACL
    Proceedings of the First Workshop on Interactive Learning for Natural Language Processing
    InterNLP Workshop @ ACL, 2021
  3. ICAPS
    Human-guided Collaborative Problem Solving: A Natural Language based Framework
    ICAPS, 2021 · demos
  4. EMNLP
    On the Effects of Transformer Size on In- and Out-of-Domain Calibration
    Findings of EMNLP, 2021
  5. EMNLP
    Few-Shot Novel Concept Learning for Semantic Parsing
    Findings of EMNLP, 2021
  6. EMNLP
    Compositional Data and Task Augmentation for Instruction Following
    Findings of EMNLP, 2021
  7. ECML-PKDD
    Variance Reduced Stochastic Proximal Algorithm for AUC Maximization
    ECML-PKDD, 2021
  8. ECML-PKDD
    Learning from Noisy Similar and Dissimilar Data
    ECML-PKDD, 2021

2020

  1. LREC
    Understanding Spatial Relations through Multiple Modalities
    LREC, 2020
  2. LREC
    From Spatial Relations to Spatial Configurations
    LREC, 2020
  3. ICML
    Goodness-of-Fit Tests for Inhomogeneous Random Graphs
    ICML, 2020
  4. COLING
    A Locally Linear Procedure for Word Translation
    COLING, 2020

2017

  1. Patent
    Techniques for generating a psychographic profile
    US Patent US20170270544A1, 2017 · Patent
  2. WOSP
    AppTechMiner: Mining Applications and Techniques from Scientific Articles
    WOSP, 2017
  3. ACL
    Towards problem solving agents that communicate and learn
    Workshop on Language Grounding for Robotics @ ACL, 2017

2016

  1. ICDM
    Identifying and characterizing truck stops from GPS data
    ICDM, 2016

2015

  1. IEEE ITSC
    Segmenting Highway Network Based on Speed Profiles
    IEEE ITSC, 2015

Curriculum Vitae

Education

  • B.Tech. in Computer Science and Engineering, Indian Institute of Technology (IIT) Kharagpur
  • Graduate study in Computer Science, University of Illinois Urbana-Champaign
  • Ph.D. in Computer and Information Science, University of Pennsylvania

Work experience

  • February 2026 – Present · Machine Learning Research Scientist, Scale AI, New York City
  • September 2024 – January 2026 · Senior Research Scientist, Microsoft, New York City
  • September 2022 – 2024 · Research Scientist, IBM Research, Yorktown Heights
  • Summer 2019 · Research Intern, RIKEN AIP, Tokyo (Prof. Masashi Sugiyama)
  • Summer 2017 · Research Intern, IBM Thomas J. Watson Research Center, Yorktown Heights (Prof. Michael Witbrock, Jie Chen, Ryan Musa)
  • Summer 2015 · Research Intern, Big Data Experience Lab, Adobe Research, Bangalore (Prof. Kokil Jaidka)
  • Summer 2014 · Research Intern, R.C. Bose Center for Cryptology and Security, Indian Statistical Institute, Kolkata (Prof. Sourav Sen Gupta)

Teaching experience

  • Teaching Assistant, CIS 419/519 (Spring 2018) — Applied Machine Learning. Instructor: Prof. Dan Roth.
  • Teaching Assistant, CIS 620 (Spring 2021) — Learning in Few-Labels Settings. Instructor: Prof. Dan Roth.

Awards and honours

  • Andrew and Shana Laursen Fellowship, University of Illinois at Urbana–Champaign, 2016.
  • President of India Gold Medal, Indian Institute of Technology Kharagpur, 2016.
  • Goralal Syngal Scholarship and Bigyan Sinha Memorial Prize, IIT Kharagpur, 2016.
  • Kishore Vaigyanik Protsahana Yojana (KVPY) fellowship, National Rank 16.