Kowndinya Renduchintala

I am a Research Associate at Adobe, in the Media and Data Science Research (MDSR) lab where I am currently looking into various things like mechanistic interpretability in the context of instruction tuning, coming up with novel pre-training procedures for LMs inspired from inductive biases that enable humans to acquire language so efficiently, and also scaling submodular optimization techniques to web-scale data.

Prior to joining Adobe, I graduated from the Indian Institute of Technology Bombay (IIT Bombay) in 2023, with a Bachelors degree (with Honors) in Computer Science and Engineering. I feel incredibly privileged to have been advised by Prof. Ganesh Ramakrishnan and Prof. Rishabh Iyer for my Undergraduate Thesis on Data-Efficient Pretraining of Language models using Submodular Functions, which I presented at EMNLP 2023.

My long-term research goal is to develop provably beneficial intelligent computational agents that can excel at both formal linguistic competence (the knowledge of rules and statistical regularities of language) and functional linguistic competence (the ability to use language in real-world situations).

If you share an interest in these topics and would like to exchange research ideas, please do not hesitate to contact me via e-mail.