Rahul Chalamala  — Resume

Graduate-level technical coursework:

• Machine Learning and Data Mining (CS 155)
• Large Language and Vision Models (EE 148)
• Learning Systems (CS 156a)
• Probability Models (ACM 116)
• Game Theory (PS 172)
• Econometrics (Ec 122)

AI/ML Researcher at Together AI

August 2023 – Current • San Francisco, CA • advised by Prof. Ce Zhang

• Reducing inference latency in large language models by researching and implementing efficient model hot-swapping algorithms in C++ and CUDA.
• Conducting experiments with knowledge distillation techniques to enable longer context windows in large multimodal models in PyTorch.

Undergraduate Researcher at Caltech Anima AI + Science Laboratory

February 2023 – October 2023 • Pasadena, CA • advised by Dr. Kaiyu Yang • Prof. Anima Anandkumar

• Explored the use of Code Large Language Models, GPT-4 and ChatGPT Plugins in Python to enhance Lean-based automated theorem provers.
• Leveraged retrieval-augmented generation to result in a 39% increase in proof success rate compared to a retrieval-free baseline model.
• Accepted for a poster presentation at the ICML 2023 Workshop on Knowledge and Logical Reasoning in the Era of Data-driven Learning.

Undergraduate Researcher at Caltech Autonomous Robotics and Controls Laboratory

October 2022 – March 2023 • Pasadena, CA • advised by Joshua Cho • Prof. Soon-Jo Chung

• Extended ArduPilot for compatibility with custom 6DOF drones to support testing and simulation for the Neural-Fly autonomous vehicle project.
• Utilized MATLAB to precisely calculate the physical dynamics of any n-motor drone, enabling it to maneuver in all six degrees of freedom.

Phillips Space Scholar at Air Force Research Laboratory

June 2021 – July 2022 • Kirtland AFB, NM • advised by Dr. Joanna Hinks

• Modeled the Air Force NTS-3 satellite's signal interference in Python and validated accuracy via field tests with the Joint Navigation Warfare Center.
• Designed a sieve-based algorithm to generate navigation ranging sequences 26.15x faster than existing GPS satellite implementations.