Nishant Holla

Investigated collaborative strategies among multiple autonomous agents in a simulated football setting, aiming to score goals against a stationary opponent using a Deep Q-Network (DQN) approach.

Project Highlights:
Automated Rendering: Dynamically renders React components with real-time previews for various props and states, eliminating the need for manual setup or live servers.

Smart Test Generation: Automatically generates test cases for user interactions such as clicks, form submissions, and modals—ensuring consistent and up-to-date test coverage with minimal manual effort.

Gained hands-on experience in constructing a general-purpose computer system and developing a modern software hierarchy from the ground up.

Over the 6-week course, I built key hardware components—starting from basic logic gates to RAM and an ALU—using VHDL, and implemented an assembler for a custom assembly language in C.

Gained a strong foundation in Data Structures and Algorithms, including topics such as trees, graphs, linked lists, and bit manipulation.

The course included four competitive programming assignments, each structured as timed contests. Achieved distinction by securing an overall rank of 3 in the class.

Gained foundational knowledge of machine learning and large language models (LLMs).

For the final project, my team developed JurisDictionary — an AI-powered legal assistant chatbot capable of answering a wide range of law-related questions. The model was trained on an extensive dataset of court proceedings and constitutional texts from around the world, leveraging OpenAI’s gpt-3.5-turbo for natural language understanding. The front end was built using Streamlit’s chat framework, providing an intuitive and interactive user experience.