Hi, my name is

Other Noteworthy Projects

• Enhancing Adversarial Attacks on LLMs

  As part of CMU's 11667 Large Language Models, working with my team, our research aimed to make adversarial attacks on large language models more effective, efficient, and adaptable to various use cases. We generate adversarial data, introduce assassin LLMs and provide reasonings on how to effectively "red team" LLMs. We focus on llama, code-llama and vicuna.

• In house Search Engine

  As part of CMU's 11642 Search Engines, I built a search engine that indexes 400M documents and implements various ranking algorithms - Simple boolean retrieval, BM25, Indri, LearningToRank, Bert, and LLMs. It also performs query expansions, diversifications and dense retrievals.

• Twitter ETL and Analytics

  As part of CMU's 15619 Cloud Computing, I built a highly scalable, available and fault-tolerant cloud service that helps query a 1TB tweets database in minimal latency. The project included, ETL of the data, designing schemas, determining deployment strategy, etc. This was one of my highest learning experiences in life.

• Cellular Network Speech Enhancement on Zoom

  As part of CMU's 11785, Intro to Deep Learning, we worked on enhancing transmitted speech over Zoom. We generate datasets containing background and transmission noise and finetune SOTA models over it. We beat industrial performance and achieve 1.92 PESQ and 0.88 STOI, as well as superior acoustic fidelity, perceptual quality, and intelligibility in various metrics.