This is Reshinth Adithyan. This is Reshinth Adithyan, you can call me Reshinth. I am a Researcher/Research Engineer working in the intersection of Machine Learning for Code Representation Learning for Developer Productivity and Static Analysis. I started liking the process of writing a parser recently. As of 2022, I am focussed on Research of Neural Program Synthesis with DSLs at Saama Research Lab and spending my free time at Code.AI which is now under Carper.AI thinking about Machine Learning applied on code.

Projects with Code Representation Learning

• gpt-code-clippy - was part of the core team in buiding the open source GitHub CoPilot during the huggingface flax/jax sprint.
• Code Representation Learningis a collective of a Machine Learning on Code tasks, which are ready to showcase using HuggingFace Spaces.
• Type Inference using AutoRegressive Pre-Trained Transformers A research project on Finetuning the codeT5 Model to infer the type of identifier in a given code snippet.
• Flash Fill Program Synthesis Interpreted IO based program synthesis using Pre-Trained Transformers.

  Have a look at my Research Proposal

  Intrinsic Evaluation for Models trained on Code Examining how well models trained with CausalLM objective, knows the context of the code.

Research Interests

• Code Representation Learning
• Parsers and Compilers
• Geometric Representation Learning in DFG(s)/CFG(s)
• Naturalness in Code
• Neural Program Synthesis
• Domain Specific Language Engineering
• Human Computer Interaction via Compilers
• Creative Coding.</strong></li>

Tools