🔱 CS @ UC San Diego 🔱
Web Developer/ ML Enthusiast
I'm Amit Namburi, a Senior at UC San Diego majoring in Computer Science. My research focuses in the field of Multimodal AI and machine's music understanding in natural language, and imagine visuals. An important part of my research is in representation learning, particularly in bridging the semantic gap between music and multi-modal media. With McAuley Lab, I'm currently delving deeper into large LLMs and fine-tuning them to cater specific needs of either better music understanding or video visual prompting. I am passionate about leveraging new technologies to address complex challenges and drive innovation in the field of AI.
Experience
McAuley Lab
Multimodal AI Research Assistant - San Diego, CA | Jan 2024 – Present
Generative Models | Large Language Models | Representation Learning
- Multimodal AI (FUTGA Model): Created and fine-tuned a generative model for fine-grained, time-aware music understanding using Large Language Models and generative augmentation techniques.
- Helped with LoRa finetuning of the SALMONN-7B backbone model on 2 NVIDIA RTX A6000 GPU.
- Conducted extensive evaluations using BLEU, METEOR, ROUGE, and BERT-score metrics, validating the effectiveness of the proposed model.
- LLM Finetuning (Video Visual Prompting): Working on fine-tuning VTime LLM model to improve on Video Visual Prompting techniques.
- Building Proactive Recommendation Systems for Netflix.
Fair Data Informatics (FDI) Lab
Student Software Engineer - San Diego, CA | March 2023 – Present
Angular.js | Pandas | NumPy | Scikit-learn
- Redesigning Foundry Dashboard architecture, a scalable ETL system for biomedical curators, with Angular.js, achieving a 40% increase in processing speed.
- Working on developing automated resource ingestion protocol for different clients hosted on EC2.
- Researched rigor and reproducibility of published work by analyzing tool performance using Pandas, NumPy, and Scikit-learn.
Computer Science and Engineering @ UC San Diego
Instructional Assistant (IA/Tutor) - San Diego, CA | April 2023 – Present
Advanced Data Structures | Tutoring
- Tutoring for upper division Advanced Data Structures course (CSE 100) with an approval rate of 100%.
- Holding 8 tutor hours weekly for a class size of 500 and accepting approximately 350 tickets.
Orektic Solutions LLP
Software Engineer Intern - Remote | June 2023 – Sep 2023
TypeScript | Cucumber.js | Selenium Driver | REST APIs | Cypress
- Built a UI Testing framework for PWR-Apps using TypeScript, Cucumber.js, and Selenium Driver.
- Worked with REST APIs using Axios library to retrieve User Token, Information, and Instance details.
- Automated Test Suite runs using Cypress which increased efficiency by 42%.
- Performed cross-browser testing with automatic report generation in Microsoft Teams and Jira to track bugs.
IPMD, Inc.
Front-End Developer Intern (Remote) - Berkeley, CA | July 2022 – Sep 2022
React.js | Typescript | Node.js
- Developed a comprehensive web application for Project M, utilizing React.js to enhance UI/UX.
- Designed and implemented dynamic, responsive webpages using Typescript and Node.js.
Publications
- Futga: Towards Fine-grained Music Understanding through Temporally-enhanced Generative Augmentation
- CoLLAP: Contrastive Long-form Language-Audio Pretraining with Musical Temporal Structure Augmentation
- CV - Visual Prompting Paper
- Proactive Recommendation Systems Project for Netflix
Projects
A website that tells your fortune based on your inner "SixthSense".
A project that detects percentages of "Happy" and "Sad" emotions and creates a playlist of 10 songs on the Spotify app.
A music recommendation and summarization system with features like music playlist generation, adding to queue based on the current song, and summarization.