Lakshya A Agrawal
Berkeley, California, United States
Summary
🤩
RockstarLakshya Agrawal is an AI PhD student at UC Berkeley and an OlMo research intern focused on making code generation by large language models more reliable and practical for real-world software engineering. With a decade of experience spanning research roles at Microsoft Research, EPFL, and IIIT-D, he blends systems-level expertise—language runtimes, IDE/debugger tooling, and transpilers—with cutting-edge work on long-context reasoning, tool use, and decoding strategies for LLMs. His Monitor-Guided Decoding (NeurIPS'23) and open-source projects like multilspy and GEPA reflect a bias toward engineering solutions that let even smaller models produce correct code at repository scale. He has a track record of shipping lasting artifacts (e.g., a Maxima→Python transpiler now in Maxima) and thrives at the intersection of programming languages, developer tooling, and generative AI.
10 years of coding experience
Stackoverflow
33reputation
909reached
1answer
0questions
Github Skills (200)
parser10
language-model10
python10
abstract-syntax-tree10
code-completion10
language-server-protocol10
metaprogramming10
java-module10
code-analysis10
code-generation10
eclipse10
static-analysis10
spoon10
codegen10
ai10
Programming languages (21)
Github contributions (5)
microsoft/multilspy
Aug 2024 - Apr 2026
multilspy is a lsp client library in Python intended to be used to build applications around language servers.
Contributions:14 releases, 60 reviews, 73 PRs in 1 year 8 months
language-serverpythonaiai4codeartificial-intelligence
LakshyAAAgrawal/CoVid
Mar 2020 - May 2020
Software for conducting live lectures and recording to as small file size as possible
Contributions:64 commits, 21 PRs, 25 pushes in 1 month
file-sizepythonconductingrecording
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.