Profile banner
Rattandeep Singh

Rattandeep Singh

Machine Learning Engineer, ex-Senior Software Engineer at Intuit

San Francisco Bay Area
New York, NY
netlify+1
Joined February 2025
Favicon imagelinkedin.com
Favicon imagerattandeep.netlify.app
Favicon imagegithub.com
Favicon imageleetcode.com

Network

9K connections
๐Ÿ’ป
Intuit Software Engineers
๐Ÿ“ˆ
Intuit Engineering & AI Leaders
๐Ÿค
Tech Talent & Product Leaders
๐ŸŽฏ
Talent Acquisition Professionals
๐Ÿ“ฑ
...
๐ŸŒŸ
...
๐Ÿง 
AI/Data Solutions

Summary

Rattandeep is a Machine Learning Engineer at Meta, specializing in AI/ML research, with a strong focus on NLP, Computer Vision, LLMs, and Generative AI. His academic pursuits at Columbia University involve cutting-edge research in language modeling and multimodal systems, contributing to significant projects like the TLDR Reddit Social Dataset. netlify+1
He possesses extensive experience in software development, particularly at Intuit, where he progressed from Software Engineer to Senior Software Engineer. His contributions spanned critical projects like migrating databases, revamping backend applications, and developing intelligent solutions using OCR, PII Guardrails, and Gen AI for products like Its Deductible and TurboTax. netlify+1
Rattandeep has a strong background in mobile application development, especially with iOS. His work at Intuit Mint involved widgetization, architectural redesigns, implementing design system libraries, and optimizing build and load times, demonstrating proficiency in creating efficient and user-friendly mobile experiences. netlify+1
His academic engagement extends to teaching, having served as a Head Graduate Teaching Assistant at Columbia University for courses including Cloud Computing and Introduction to Databases. He is committed to continuous learning and sharing knowledge, evident from his leadership in guiding and mentoring students. netlify+2
Rattandeep is a contributor to academic research, co-authoring papers on program synthesis dialog agents and benchmarks for form completion using vision-language agents. This showcases his involvement in advancing the theoretical and practical aspects of AI and natural language processing. arxiv+1

Work

Education

Projects

Writing

FormGym: Doing Paperwork with Agents

June 1, 2025

A research paper proposing FormGym, a new benchmark for evaluating the ability of general-purpose vision-language agents (VLAs) to perform end-to-end form completion in pure-image domains. It also introduces FieldFinder, a tool that significantly improves VLA accuracy by assisting with identifying where to place text on a form.

Favicon imagearxiv.org

Program Synthesis Dialog Agents for Interactive Decision-Making

February 1, 2025

A research paper introducing BeNYfits, a novel benchmark for evaluating agents on public benefits eligibility decisions, and ProADA, an approach utilizing program synthesis for interactive decision-making. ProADA aims to improve decision-making by mapping dialog planning to a code generation problem and intelligently using structured data gaps.

Favicon imagearxiv.org