Welcome!

About me

I’m a Computer Science Ph.D. student at the University of Texas at Dallas. My current research is on automated logical reasoning in natural language. I love building applications that simplify human lives. In my free time, you’ll find me playing guitar, reading a book or on a climbing wall.

Projects

Improving Natural Language Reasoning using ASP and LLMs (ICLP ‘23)

STAR (Semantic parsing Transformer and Asp Reasoner) framework is applied to reasoning challenges such as qualitative reasoning (reasoning about physical properties), mathematical reasoning or goal-directed conversation (a bot that helps you find a restaurant). By converting text to logic predicates using Large Language Models (GPT-3) and reasoning over them using commonsense knowledge encoded using Answer Set Programs, we demonstrate an improvement over baseline models. Accepted in the ICLP ‘23 conference.

Image to Rhyming Poetry

Given an image caption as stimulus, poetry is generated. The challenge is making the poetry rhyme, fluent (less repetition) and relevant to the caption content at the same time. The outputs of different models for the same image of a flower (caption: ‘A closeup of a flower’) are given below:

Reinforcement Learning with Human Feedback

Pacman learns using reinforcement learning (Function approx. Q learning) improved through human feedback. After training, a test run looks like:

Movie recommendation webapp with CI/CD

A website which recommends a movie using content-based recommendation algorithm. It is setup using continuous integration, deployment and monitoring. (Uses Jenkins, Docker, Rundeck and Logstash-Elastic-Kibana stack)

Please see my Github and LinkedIn profiles for other projects.

Skills

Data Science: Pytorch, Scikit-Learn, Pandas, NumPy, Matplotlib, OpenCV
Programming: Python, Java, C++, C
Software Development: Flask, Spring, Docker, Jenkins, Bash
Languages: English, Telugu, Hindi, Malayalam

Page template forked from evanca