Spotlight
Conference on Computer Vision and Pattern Recognition (CVPR)
• 2026
Same or Not? Enhancing Visual Perception in Vision-Language Models
Damiano Marsili , Aditya Mehta , Ryan Y. Lin , Georgia Gkioxari
Ryan Y. Lin
AI Researcher in Quant Finance @ DRW |
prev. ML @ Apple
M.S. CS @ Stanford | Caltech Alum
Scroll
About
Hi, I'm Ryan.
Palo Alto, CA
I'm currently conducting cutting-edge AI research in quantitative finance at DRW while completing a part-time M.S. in CS at Stanford. Before this, I completed my undergraduate B.S. at Caltech, where I majored in Computer Science and was also a part of the NCAA DIII Varsity Swim & Dive team.
During my three years at Caltech, I also had the privilege of researching and working under some amazing professors! Big thanks to Professors Adam Blank, Adam Wierman, Anima Anandkumar, Georgia Gkioxari, and Yisong Yue for mentoring me in research projects and when I was a TA for their courses.
I spent one summer working on optimizing model inference at Apple, Inc. I spent the other as a full time researcher in Prof. Anandkumar's lab as a Summer Undergraduate Research Fellow. Throughout that time, I really gained a love for going to the beach and cooking (thanks, Tom Mannion)!
In my spare time, I enjoy swimming, playing basketball, lifting weights, playing video games, and making ice cream :)
Research
Research Interests
Exploring continuous physical systems, geometric representations, quantitative finance, and visual perception.
01 Caltech Anima Lab
Physics-Informed ML
Continuous operators & partial differential equations
Modeling infinite-dimensional continuous operators across arbitrary physical geometries. Enabling zero-shot resolution-invariant surrogate models for fluid flows, wave dynamics, and 3D MRI reconstruction.
Focus:
mGNO (TMLR '25) • 3D MRI (NeurIPS '25)
02 Personal Research @ Caltech
Representation Learning
Geometric deep learning & manifold curvature
Investigating how neural representations adapt to complex non-Euclidean geometry. Designing curvature-adaptive architectures and diffusion models that respect intrinsic data manifolds.
Focus:
CAT (NeurIPS '25) • Graph Diffusion (NeurIPS '25)
03 Personal Research @ Caltech
Deep Learning in Finance
Quantitative modeling & high-frequency dynamics
Applying modern deep learning architectures and large models to financial systems or game-theoretic simulations.
Focus:
Strategic Collusion in LLMs (NeurIPS '24)
04 Caltech Gkioxari Lab
Computer Vision
Perceptual grounding & vision-language models
Probing visual discrimination in multimodal foundation models. Developing rigorous counterfactual benchmarks to evaluate whether vision-language models genuinely perceive visual distinctions.
Focus:
Same or Not? (CVPR '26 Spotlight)
Publications
Selected Works
NeurIPS Workshop on Non-Euclidean Foundation Models and Geometric Learning
• 2025
CAT: Curvature-Adaptive Transformers for Geometry-Aware Learning
Ryan Y. Lin , Siddhartha Ojha , Nicholas Bai
NeurIPS Workshop on Imageomics: Discovering Biological Knowledge from Images Using AI
• 2025
Coarse-to-Fine 3D MRI Reconstruction via 3D Neural Operators
Armeet Singh Jatyani , Jiayun Wang , Ryan Y. Lin , Valentin Duruisseaux , Anima Anandkumar
NeurIPS Workshop on New Perspectives in Graph Machine Learning
• 2025
Diffusion-Generated Social Graphs Enhance Bot Detection
Alec Laprevotte , Ryan Y. Lin , Siddhartha Ojha
Transactions on Machine Learning Research (TMLR)
• 2025
Enabling Automatic Differentiation with Mollified Graph Neural Operators
Ryan Y. Lin , Julius Berner , Valentin Duruisseaux , David Pitt , Daniel Leibovici , Jean Kossaifi , Kamyar Azizzadenesheli , Anima Anandkumar
NeurIPS Workshop on Language Gamification
• 2024
Strategic Collusion of LLM Agents: Market Division in Multi-Commodity Competitions
Ryan Y. Lin , Siddhartha Ojha , Kevin Cai , Maxwell F. Chen
Experience
My Journey
A blend of research, traditional software engineering, and community building across academia and industry.
2026 — Present
AI Researcher at DRW Holdings
Conducting AI/ML research in the domain of quantitative finance.
#Quantitative ML
#Deep Learning
#LLMs
2026 - Present
M.S. in Computer Science at Stanford University
Pursuing graduate studies under the Honors Cooperative Program (HCP), focusing on advanced foundation models, geometric deep learning, and continuous mathematical representations.
#Stanford CS
#Graduate Studies
2023 — 2026
B.S. in Computer Science at California Institute of Technology (Caltech)
Undergraduate researcher at Anima Lab, Yue Lab, and Gkioxari Lab. Executive Director of Hacktech (Caltech's annual hackathon). NCAA Division III Varsity Swimmer (top 10 all-time!). Teaching Assistant for undergraduate + graduate computer science courses, and Cooking Basics (SA 16).
#Caltech CS
#Hacktech Director
#NCAA
#TA
Summer 2025
AI/ML Engineering Intern at Apple (Foundation Model Compute)
Designed, implemented, and deployed a production-grade LLM inference server from scratch with elastic accelerator autoscaling, improving resource efficiency by 8x. Selected to present to Apple AI/ML leadership. Contributed to axlearn, Apple's largest open-source AI framework.
#axlearn
#LLM Inference
#Autoscaling
2021 — 2024
Software Engineer Intern (4 Summers) at The MITRE Corporation
Designed a pipeline using Retrieval-Augmented Generation (RAG) and LLMs to automate security profile generation from STIGs, accelerating delivery by 500%. Built automated cloud pipelines on AWS EC2. Developed large, open-source cybersecurity data normalization and visualization platforms for corporate partners and government sponsors.
#RAG
#Cybersecurity
#DevSecOps
#AWS
3
Erdős Number
via Ryan Lin → Anima Anandkumar → Béla Bollobás →
Paul Erdős
NCAA DIII
Varsity Swimmer
Caltech Men's Swimming & Diving
Languages
Fluent in:
English, Chinese, Burmese
Projects
Things I've Built
Interactive tools and mathematics exploring audio separation, spatial mapping, and geometric proofs.
Audio Machine Learning
AI Vocal Remover
A client-side web application that isolates instrumental backtracks from audio files in real time using Demucs inference and the Web Audio API.
Web Audio API Demucs ML Client-Side Audio
Spatial Mapping
Archived
Trip Restaurant Planner
An interactive mapping tool for discovering and saving places with background geocoding, spatial list views, and Leaflet mapping. (Archived / No longer serviced).
Archived Leaflet GIS Spatial Indexing
Research Tool
Academic Paper Reader
A peaceful, distraction-free PDF reading environment tailored for reading machine learning papers, featuring personal annotations and sidebar indexing.
PDF.js Research Workflow Tailwind
Pure Mathematics
Lattice Point Geometry Portfolio
A collection of 100+ formal proofs exploring fundamental geometric, algebraic, and topological properties of discrete lattice planes.
Discrete Geometry Formal Proofs Mathematics
Let's Connect
Whether you'd like to collaborate on research, talk about mathematics and physics, or discuss basketball, swimming and cooking, I'd love to hear from you.
Copied to clipboard