Rohan Bavishi
Member of Technical Staff
Amazon AGI
rohan.bavishi95@gmail.com
rbavishi@cs.berkeley.edu
@code_monet
@code_monet
rbavishi
Google Scholar
CV
At Amazon AGI Autonomy (previously Adept AI Labs), I have previously worked across the modeling stack – architecture and pretraining for the Fuyu-8B and Fuyu-Heavy models, agent post-training for Adept Experiments, Nova Act – as well as product engineering wherein I developed the Adept Workflow Language (AWL).
In my past life, I completed my Ph.D. in Programming Languages and Systems at the University of California, Berkeley, where I was advised by Koushik Sen. I was also a member of the Programming Systems group. My research focused on designing tools and techniques for improving the productivity of programmers, specifically data scientists. In addition, I’ve researched techniques for automated program repair for static analysis violations and syntax errors in formula languages.
I obtained my bachelor’s degree in computer science from the Indian Institute of Technology, Kanpur where my undergraduate research was focused on developing more precise bug localization techniques.
Feel free to contact me via email, or any of the platforms above. Cheers!
-
Tools and Techniques for Building Programming Assistants for Data Analysis
Ph.D. Thesis, UC Berkeley, 2022
-
Neural-Backed Generators for Program Synthesis
Master's Thesis, UC Berkeley, 2019
-
Useful, reliable agents from prototype to production*
2025
-
Introducing Amazon Nova Act*
2025
-
Building Powerful Agents with Adept*
Adept AI Labs, 2024
-
Adept Fuyu-Heavy: A new multimodal model
Adept AI Labs, 2024
-
Introducing Adept Experiments
Adept AI Labs, 2023
-
Fuyu-8B: A Multimodal Architecture for AI Agents
, , , , , , and
Adept AI Labs, 2023
-
Neurosymbolic repair for low-code formula languages
, , , , , , , and
Proceedings of the ACM on Programming Languages (OOPSLA) 2022
-
VizSmith: Automated Visualization Synthesis by Mining Data-Science Notebooks
, , , , and
Proceedings of the 36th IEEE/ACM International Conference on Automated Software Engineering (ASE) 2021
-
Gauss: Program Synthesis by Reasoning over Graphs
, , , and
ACM International Conference on Object Oriented Programming Systems Languages and Applications (OOPSLA) 2021
-
AutoPandas: Neural-Backed Generators for Program Synthesis
, , , , and
ACM International Conference on Object Oriented Programming Systems Languages and Applications (OOPSLA) 2019
-
Phoenix: Data-Driven Synthesis of Repairs for Static Analysis Violations
, , and
ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (ESEC/FSE) 2019
* Code not available due to proprietary reasons -
To Be Precise : Regression Aware Debugging
, , and
ACM International Conference on Object Oriented Programming Systems Languages and Applications (OOPSLA) 2016
-
Phoenix: A Tool for Automated Data-Driven Synthesis of Repairs for Static Analysis Violations
, , , , , and
ACM/IEEE 42nd International Conference on Software Engineering (ICSE) 2020
* Code not available due to proprietary reasons
-
Neural Inference of API Functions from Input–Output Examples
, , , , , and
Workshop on ML for Systems at NeurIPS 2018
-
Regression Aware Debugging for Mobile Applications
, , and
International Workshop on Mobile Development (Mobile!) 2016
-
Context2Name: A Deep Learning-Based Approach to Infer Natural Variable Names from Usage Contexts
, , and
2018
-
Language-agnostic computer program repair engine generator
, , , , , , , , and
2022
-
Data-Driven Synthesis of Fix Patterns
, , and
2020
Reviewer
- Program Committee, OOPSLA 2025
- Program Committee, SPLASH 2025 Student Research Competition
- Artifact Evaluation Committee, OOPSLA 2020
- Artifact Evaluation Committee, PLDI 2020
Subreviewer
- PLDI 2022
- ICSE 2022
- POPL 2021
- ISSTA 2020
- ICST 2020
- ASE 2018
- CAV 2018
- ASPLOS 2018
- PLDI 2018