Hi, I'm Ansh!

I'm a final semester senior CS student at Rutgers University New Brunswick, based in central New Jersey.

My coursework spans from basics like data structures and algorithms to specifics like software methodology, intro to artificial intelligence, data science topics, and programming language theory. I've studied in languages like Java, C, and Python, taken courses using OCaml and R, as well as used popular libraries like NumPy and Pandas. Most of my personal work now is in applied AI and backend work in Python.

What I'm Looking For

I'm looking for full time new grad software engineering roles starting in January 2027, particularly in applied AI and backend development.

I'm open to work within the continental US, but particularly targeting the NYC metro area. I'm currently based in central NJ.

Recent Work

Most recently, I built a music listening history and analytics tool because I felt Spotify's history feature did not give me enough information. For instance, history from many weeks ago, or listening trends are completely absent, so I implemented them myself. I chose to use Last.fm's API to source scrobbles because it's easier to work with and can encompass streams from multiple services under one account's umbrella. You can demo it yourself if you've signed up for Last.fm by visiting rotation.ajetli.com, and read more about it at the linked GitHub repo.

Before that I worked on an agentic RAG pipeline over 1000 research papers spanning topics in AI, which I personally use as a learning tool. I chose RAG rather than a normal LLM call not only to learn the technology but because it helps avoid hallucination and reliance on training data. The knowledge base can also be quickly and cheaply expanded in the future. You can ask it any question on AI/ML related topics, and it will only formulate an answer based on its source papers, which it can cite and source for you. It works on a LangGraph agent that breaks queries into subqueries, fetches relevant chunks, reranks them, and checks if a sufficient answer can be formed. If not, it tries again. You can demo it yourself at rag.ajetli.com, or read more about it under the architecture tab. You can also view all the source papers under the papers tab.

An ongoing project of mine is a presently unnamed text based RPG game where an LLM acts as a world builder and narrator, almost like a dungeon master. The idea was to be able to play through your own book, where narratively you were the main character but you did not dictate the state of the world beyond. You are free to make your own actions, and the world should respond in kind. However, to make this work as a full scale project I had to figure out logging and retrieval strategies. My solution uses a deterministic Python engine that owns the game state and an LLM that returns game state alterations in an enforced structured output format, allowing the engine to receive and resolve all changes and reduce risk of hallucination. I'm constantly iterating on this project and you can read more about it or look through the code at the GitHub repo linked above.

Education

Hobbies

When I'm not busy with school or projects, I enjoy reading and playing video games. I'm a fan of anything in the fantasy genre, and right now I'm playing a lot of indie games. I'm a long time fan of multiplayer games like Destiny 2 and Overwatch. I'm also really into music, and I'm learning how to play the guitar!

Contact

The best way to reach me is via email or LinkedIn. Please feel free to reach out.