Akhilesh Gotmare

I am an AI researcher based in San Francisco. Most recently, I was a Sr Researcher with Salesforce's AI Research org where I worked with Richard Socher, Caiming Xiong, and Steven Hoi.

My research spans code generation with RL (CodeRL, NeurIPS '22), perf-aware code optimization (PerfCodeGen, ACM SIGSOFT '25), open code LLMs (CodeT5+, EMNLP '23), and controllable text generation (GeDi, EMNLP '21). On the applied front, I led research efforts for ApexGuru - Salesforce's code review agent now used by ~10k enterprise orgs and featured at Dreamforce (dev keynote) for three consecutive years ('23, '24, '25).

I was trained at EPFL (Master's in CS) where I worked with Martin Jaggi, and prior to that I completed my undergrad in EE at IIT.

Email  |  Google Scholar  |  LinkedIn  |  GitHub

Publications
PerfCodeGen: Improving Performance of LLM Generated Code with Execution Feedback
Yun Peng, Akhilesh Gotmare, Michael Lyu, Caiming Xiong, Silvio Savarese, Doyen Sahoo
FORGE 2025 (ACM SIGSOFT Distinguished Paper Award)
[ arXiv, FORGE '25, Code ]

CodeT5+: Open Code LLMs for Code Understanding and Generation
Yue Wang, Hung Le, Akhilesh Gotmare, Nghi Bui, Junnan Li, Steven Hoi
EMNLP 2023
[ arXiv, Blog, Code ]

CodeRL: Mastering Code Generation through Pretrained Models and Deep Reinforcement Learning
Hung Le, Yue Wang, Akhilesh Gotmare, Silvio Savarese, Steven Hoi
NeurIPS 2022
[ arXiv, Blog, Code, Media ]

Cascaded Fast and Slow Models for Efficient Semantic Code Search
Akhilesh Gotmare, Junnan Li, Shafiq Joty, Steven Hoi
Foundations of Software Engineering 2023
[ FSE 2023 ]

Align before fuse: Vision and language representation learning with momentum distillation
Junnan Li, Ramprasaath Selvaraju, Akhilesh Gotmare, Shafiq Joty, Caiming Xiong, Steven Hoi
NeurIPS 2021 (Spotlight)
[ arXiv, Blog, Code ]

GeDi: Generative Discriminator Guided Sequence Generation
Ben Krause*, Akhilesh Gotmare*, Bryan McCann, Nitish Shirish Keskar, Shafiq Joty, Richard Socher, Nazneen Fatema Rajani
EMNLP 2021 (Findings)
[ arXiv, Blog, VB coverage, Import AI, Code, Demo ]

A Closer Look at Deep Learning Heuristics: Learning rate restarts, Warmup and Distillation
Akhilesh Gotmare, Nitish Shirish Keskar, Caiming Xiong and Richard Socher
ICLR 2019 (partial results presented at CRACT Workshop, NeurIPS 2018)
[ arXiv, OpenReview, Poster ]

Using Mode Connectivity for Loss Landscape Analysis
Akhilesh Gotmare, Nitish Shirish Keskar, Caiming Xiong and Richard Socher
Workshop on Modern Trends in Nonconvex Optimization for Machine Learning, ICML 2018.
[ arXiv, Poster ]
Decoupling Backpropagation using Constrained Optimization Methods
Akhilesh Gotmare*, Valentin Thomas*, Johanni Brea and Martin Jaggi
Workshop on Efficient Credit Assignment in Deep Learning and Deep Reinforement Learning, ICML 2018.
[ OpenReview, Poster ]

Unsupervised robust nonparametric learning of hidden community properties
Mikhail Langovoy, Akhilesh Gotmare, Martin Jaggi, Suvrit Sra
Mathematical Foundations of Computing 2 (2), 127-147
[ arXiv, AIMS MFC journal ]

Swarm and evolutionary computing algorithms for system identification and filter design: A comprehensive review
Akhilesh Gotmare, Sankha Subhra Bhattacharjee, Rohan Patidar, Nithin V. George
Swarm and Evolutionary Computation, 32, 68-84.
[ Elsevier SWEC journal ]

Nonlinear system identification using a cuckoo search optimized adaptive Hammerstein model
Akhilesh Gotmare, Rohan Patidar, Nithin V. George
Expert systems with applications, 42(5), 2538-2546.
[ Elsevier ESWA journal ]

Technical Blogs
MCP+: Precision Context Management for MCP Agents
Prathyusha Jwalapuram, Akhilesh Gotmare, Doyen Sahoo, Silvio Savarese, Junnan Li
MCP+ project page, 2026
[ Blog, Code ]

How Amazon Bedrock Custom Model Import streamlined LLM deployment for Salesforce
Srikanta Prasad, Utkarsh Arora, Raghav Tanaji, Nitin Surya, Gokulakrishnan Gopalakrishnan,
Akhilesh Gotmare, Hrushikesh Gangur
AWS Machine Learning Blog, Oct '25
[ Blog ]

The ApexGuru AI Engine, Explained by Mayuresh Verma, Akhilesh Gotmare
Salesforce Developers Blog, June '25
[ Blog ]


Website style cloned from Jon Barron's website