publications and projects

Publications and research projects in reversed chronological order. * denotes equal contribution.

2026

  1. ICLR
    futurefill.png
    FutureFill: Fast Generation from Convolutional Sequence Models
    Naman Agarwal, Xinyi Chen, Evan Dogariu, Devan Shah , and 5 more authors
    ICLR 2026
    arXiv Code Website

2025

  1. NeurIPS
    spectralds.png
    SpectraLDS: Provable Distillation for Linear Dynamical Systems
    Devan Shah, Shlomo Fortgang, Sofiia Druchyna, and Elad Hazan
    NeurIPS 2025
    arXiv Code Slides Website
  2. NeurIPS-W
    upskill.png
    UpSkill: Mutual Information Skill Learning for Structured Response Diversity in LLMs
    Devan Shah*, Owen Yang*, Daniel Yang, Chongyi Zheng , and 1 more author
    NeurIPS 2025 Workshop on Scaling Environments for Agents
    arXiv Code Website
  3. IEEE TMC
    scooterid.png
    ScooterId: Posture-Based Continuous User Identification from Mobility Scooter Rides
    Devan Shah, Ruoqi Huang, Nisha Vinayaga-Sureshkanth, Tingting Chen , and 1 more author
    IEEE Transactions on Mobile Computing 2025
    PDF
  4. mi_skills.png
    Scaling Mutual Information Skill Learning to Larger Skill Spaces
    Devan Shah*, and David Shustin*
    Project
    PDF
  5. contrastive.png
    Augmenting Large Reasoning Models with Contrastive Goal-Conditioned Reinforcement Learning
    Devan Shah*, Kevin Wang*, and David Yan*
    Project
    PDF Code Poster
  6. camelax.png
    Camelax: A Lightweight Machine Learning Library in OCaml
    Devan Shah*, and Daniel Yang*
    Project
    PDF Code

2024

  1. AAAI
    aaai.png
    Rider Posture-Based Continuous Authentication with Few-Shot Learning for Mobility Scooters (Student Abstract)
    Devan Shah, Ruoqi Huang, Tingting Chen, and Murtuza Jadliwala
    HTML Poster Slides
  2. wave_filtering.png
    Wave Filtering for General Linear Dynamical Systems
    Devan Shah*, and Brandon Cho*
    Project
    PDF Code
  3. bloom_filters.png
    Truly Adaptive Bloom Filters
    Devan Shah*, and David Yan*
    Project
    PDF Poster
  4. Survey
    ssm_survey.png
    A Survey of State Space Models: From Linear Systems to Language
    Devan Shah*, and Brandon Cho*
    Survey
    PDF
  5. Survey
    deep_and_cheap.png
    Discussion on "Why Does Deep and Cheap Learning Work So Well?"
    Devan Shah
    Survey
    PDF

2023

  1. dreamscape.png
    DreamScape: Denoising CLIP Embeddings with User Images for Improved Visualization Reconstruction
    Devan Shah
    Project
    PDF Code
  2. Survey
    mwu.png
    Understanding Dynamic Algorithms for Packing-Covering LPs via Multiplicative Weight Updates
    Devan Shah*, Owen Yang*, and Sunil Vittal*
    Survey
    PDF