I am a Research Fellow at Microsoft Research India, working with Aditya Kanade and Sriram Rajamani on AI4Code. My current research is about how to augument language models with program analysis to improve how they reason about code and to automate complex software engineering tasks. Check out our most recent work in this direction: CodePlan.

Previously, I worked with Florian Shkurti at the University of Toronto on learning based approaches into task and motion planning for robotics. As an undergraduate student at BITS Pilani Goa, I was affiliated with APPCAIR where I worked with Prof. Ashwin Srinivasan and Dr Gautam Shroff on neuro-symbolic approaches to visual reasoning.

News

Jan 2024 | Both CodePlan and CORE are accepted at FSE 2024!
Dec 2023 | I'll be attending NeurIPS to present CodePlan as a poster at the FMDM Workshop.
Dec 2023 | I am applying to grad school this cycle.

Publications

CodePlan: Repository-level Coding using LLMs and Planning
Ramakrishna Bairi, Atharv Sonwane, Aditya Kanade, Vageesh D C, Arun Iyer, Suresh Parthasarathy, Sriram Rajamani, B. Ashok, Shashank Shet FMDM, NeurIPS 2023 | Workshop on Foundation Models for Decision Making
Accepted at FSE 2024 | ACM International Conference on the Foundations of Software Engineering
preprint

Frustrated with Code Quality Issues? LLMs can Help!
Nalin Wadhwa, Jui Pradhan, Atharv Sonwane, Surya Prakash Sahu, Nagarajan Natarajan, Aditya Kanade, Suresh Parthasarathy, Sriram Rajamani
Accepted at FSE 2024 | ACM International Conference on the Foundations of Software Engineering
preprint

Policy-Guided Lazy Search with Feedback for Task and Motion Planning
Mohamed Khodeir, Atharv Sonwane, Ruthrash Hari, Florian Shkurti Best Paper | Long Horizon Planning Workshop at CoRL 2022
ICRA 2023 | IEEE International Conference on Robotics and Automation
abstract| paper| website

Neural Analogical Reasoning
Atharv Sonwane, Abhinav Lalwani, Sweta Mahajan, Gautam Shroff, Lovekesh Vig NeSy 2022 | International Workshop on Neural-Symbolic Learning and Reasoning
paper

Solving Visual Analogies Using Neural Algorithmic Reasoning
Atharv Sonwane, Gautam Shroff, Lovekesh Vig, Ashwin Srinivasan, Tirtharaj Dash Oral Presentation Finalist | Student Abstract & Poster at AAAI 2022
paper

Using Program Synthesis and Inductive Logic Programming to solve Bongard Problems
Atharv Sonwane*, Sharad Chitlangia*, Tirtharaj Dash, Lovekesh Vig, Gautam Shroff, Ashwin Srinivasan (* = Equal Contribution)
AAIP 2021 | 10th International Workshop on Approaches and Applications of Inductive Programming
paper