Skip to content
View pirroh's full-sized avatar


  • Pro


@replit @snap-stanford @HumanDynamics @isi-usc-edu @eXascaleInfolab @googlers @stanford-cs329s
Block or Report

Block or report pirroh

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Please don't include any personal information such as legal names or email addresses. Maximum 100 characters, markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse

👋   Hi there, I'm Michele Catasta

👨‍💻   VP of AI at Replit (building the future of software development with AI)

🔬   Former Head of Applied Research @ Google Labs (working on AI applied to Source Code, Large Language Models)

👨‍🏫   Former Research Scientist and Instructor in AI @ Stanford University

🧐   Expertise: Large Language Models, AI for Code, Machine Learning, Information Retrieval, Data Science

🌐  [Personal page] - [CV] - [LinkedIn] - [X] - [Google Scholar]

‼️   News

When What Links
Apr 2024 Replit Code Repair announced at Replit Developer Day [tech report] - [X thread] - [video] - [media]
Oct 2023 Replit AI for All announced at AI Engineer Summit [video] - [media] - [blog post]
Jun 2023 I published the Replit AI Manifesto [blog post]
May 2023 PaLM 2 announced at Google I/O -- I worked on code pre-training and evaluations [paper] - [blog post] - [website]
May 2023 Natural Language to Code Generation in Interactive Data Science Notebooks accepted at ACL 2023 [paper]
Apr 2023 replit-code-v1-3b announced at the Replit Developer Day and released opensource [X thread] - [video] - [HuggingFace model] - [GitHub repo]
Apr 2023 Measuring the Impact of Programming Language Distribution accepted at ICML 2023 -- I was the Principal Investigator [paper] - [code]
H2 2022 Invited talks on AI meets Source Code: status quo and outlooks [video] and events: [EPFL], [Synapse AI Symposium], [Berkeley AI Summit] & more
H2 2022 PaLM: Scaling Language Modeling with Pathways submitted to the Journal of Machine Learning Research -- I worked on PaLM-Coder [paper] - [blog post]
Mar 2021 Language-Agnostic Representation Learning of Source Code from Structure and Context (AKA Code Transformer) accepted at ICLR 2021 [paper] - [demo] - [code]

🔦   Highlights

🎓   Education

👨‍💻   Experience

  • Head of Applied Research at Google X & Google Labs
    • Worked on Large Language Models and AI for Code (including PaLM and PaLM 2)
  • Research Scientist at Stanford University and at EPFL
    • Contributed to several projects (funded by IARPA, DARPA, Samsung, Google, Amazon, ...) with research on Deep Learning (GNNs, Transformers, Open Graph Benchmark, etc.), Recommender Systems, Crowdsourcing, and Data Science.
  • Intern at MIT Media Lab (w/ Prof. Alex 'Sandy' Pentland), Yahoo Research (w/ Prof. Ricardo Baeza-Yates), and Google.
  • Co-founder of, the largest Semantic Web Search Engine (back in the days). The core technologies developed for Sindice evolved into:
    • a top-level Apache project, Any23
    • several contributions to Hadoop, Lucene and Solr
    • Siren, an investigative intelligence platform which secured $15M+ in funding -- kudos to my amazing ex-colleagues 👍

👨‍🏫   Teaching

Pinned Loading

  1. replit/ReplitLM replit/ReplitLM Public

    Inference code and configs for the ReplitLM model family

    Python 914 75