Hi, I'm

Arijit Dey

Incoming Ph.D. Student at MBZUAI

I build and study machine learning systems spanning vision-language models, graph & sequence representation learning, and optimization for modern deep learning.

Download CV Publications
Arijit Dey

News

  • 2026
    ★My first paper was accepted! Graph State Networks — Persistent Nodewise Selective State Space Models was accepted to TMLR (Transactions on Machine Learning Research), 2026.  PDF · Code
  • 2026
    Joining MBZUAI as an incoming Ph.D. student in Machine Learning.
  • 2026
    Submitted Convergence of Asynchronous SGD for Polyak–Łojasiewicz Functions to IEEE Transactions on Automatic Control.  PDF · Code

What I'm working on

I like working end-to-end — from problem formulation and modeling to training infrastructure, honest evaluation, and clean open-source code.

Vision-Language for Medical Imaging

At Symviq, I built CLIP-style vision-language models for chronic kidney and heart disease classification from retinal fundus images, and foundation-style encoders for downstream tasks.

Graph & Sequence Models

Selective state-space models over graphs — my paper Graph State Networks studies persistent, nodewise state for scalable graph representation learning.

Optimization for Deep Learning

Convergence theory for asynchronous / delayed SGD under the Polyak–Łojasiewicz condition — a useful middle ground between convex and general nonconvex objectives.

What's next

Incoming · Fall 2026

Ph.D. at MBZUAI

I'm joining the Mohamed bin Zayed University of Artificial Intelligence to pursue research in optimization, multimodal and graph machine learning. Always happy to chat about collaborations.

Mohamed bin Zayed University of Artificial Intelligence campus

Research interests

Vision-language & contrastive learning Multimodal retrieval & grounding Graph representation learning Selective state-space models Optimization & convergence theory Asynchronous / distributed SGD Medical imaging

Selected publications

TMLR 2026 · First paper

Graph State Networks — Persistent Nodewise Selective State Space Models

Arijit Dey

Transactions on Machine Learning Research (TMLR), 2026

PDF Code
Under review

Convergence of Asynchronous Stochastic Gradient Descent for Polyak–Łojasiewicz Functions

Arijit Dey, Bahman Gharesifard

Submitted to IEEE Transactions on Automatic Control

PDF Code

See all publications →

Education

MS, Electrical & Computer Engineering

University of California, Los Angeles (UCLA) · 2023–2024

M.Tech, Electrical Engineering

Indian Institute of Science (IISc) · 2021–2023

B.E., Instrumentation & Electronics Engineering

Jadavpur University · 2017–2021

Want the detailed version, with milestones and links? See Research and Projects.