Devan ShahI am a rising senior in Computer Science at Princeton University pursuing a minor in mathematics, and I am grateful to be advised by Prof. Elad Hazan. Through the Hazan Lab, I work with Google DeepMind Princeton on language modeling and dynamical systems. Previously I worked with Prof. Tingting Chen on ML for authentication. Broadly, I am interested in better understanding language models, such as improving how models reason, designing post-transformer architectures, and thinking more about post-training. At Princeton, I help host a few reading groups and chair our ACM chapter. Some recent work I am especially proud of is [1] and [2]. Email / GitHub / Google Scholar / LinkedIn |
Image Snippet from PRD presentation |
Work Experience |
|
Quantitative Research Intern at Jane StreetMay 2025 — PresentI am interning as a (Quantitative) Research intern at Jane Street. |
|
Hazan Lab w/ Google DeepMind PrincetonSeptember 2024 — PresentI work on language modeling and dynamical systems in the Hazan Lab and collaborate with Google DeepMind Princeton. I've worked on projects aiming to accelerate the Google Spectral Transformer and better model linear dynamical systems, such as such as SpectraLDS, FutureFill, and Google Deluca. |
|
Recommendation Systems at TikTokMay 2024 — August 2024I improved the recommendation system for ecommerce videos for a projected $33,000,000+ increase in annual sales through projects focusing on improving multi-interest modeling. |
|
EEG Vision Embeddings at CareYayaNovember 2023 — May 2024I studied the reconstruction of viewed images from brain activity, improving performance by conditioning on a user's camera roll. We aimed to better align EEG embeddings with CLIP space. |
|
Price Prediction at Ticket Wallet (YC X25)June 2023 — August 2023I worked at an early-stage startup to advise and design data pipelines, pricing algorithms, and refine product pitches. I also worked on algorithms to determine if a ticket was likely to be sold. |
Papers & Projects |
|
SpectraLDS: Provable Distillation for Linear Dynamical SystemsDevan Shah, Shlomo Fortgang, Sofiia Druchyna, Elad Hazan Preprint paper / code / slides |
|
FutureFill: Fast Generation from Convolutional Sequence ModelsNaman Agarwal, Xinyi Chen, Evan Dogariu, Devan Shah, Hubert Strauss, Vladimir Feinberg, Daniel Suo, Peter Bartlett, Elad Hazan Preprint paper / code / slides |
|
Augmenting Large Reasoning Models with Contrastive Goal-Conditioned Reinforcement LearningDevan Shah*, Kevin Wang*, David Yan* (* indicates equal contribution) paper / code / poster |
|
Parallel Scaling with Entropic ReasonersDevan Shah*, Owen Yang*, Daniel Yang* paper / code / poster |
|
ScooterId: Posture-Based Continuous User Identification From Mobility Scooter RidesDevan Shah , Ruoqi Huang, Nisha Vinayaga-Sureshkanth, Tingting Chen, Murtuza Jadliwala IEEE TMC 2025 website / paper |
Surveys |
|
Discussion on "Why Does Deep and Cheap Learning Work So Well?"Devan Shah paper |
|
A Survey of State Space Models: From Linear Systems to LanguageDevan Shah* , Brandon Cho* (Equal Contribution) paper |
|
Understanding Dynamic Algorithms for Packing-Covering LPs via Multiplicative Weight UpdatesDevan Shah* , Owen Yang*, Sunil Vittal* paper |
Coursework |
Computer Science |
||||||||||||||||
|
||||||||||||||||
Math |
||||||||||||||||
|
Theory |
||||||||||||||
|
||||||||||||||
Creative Writing |
||||||||||||||
|
Hackathons |
|