ML Engineer @ TikTok USDS
I completed my Masters in Computer Science and Engineering from University of California, San Diego. Prior to joining UCSD, I did my BTech with Honors in Computer Science from IIT Gandhinagar. My interests lie in the areas of building technology-driven solutions for real-world applications. I am particularly intrigued by cross-disciplinary research topics and always strive to expand my knowledge and explore new areas.
I have a keen interest and have pursued projects in a diverse range of areas including Generative AI, Applied ML, Natural Language Processing, Reinforcement Learning, Data Science, Bayesian Modeling, Computer Vision. I am passionate about utilizing these fields to create pioneering solutions and actively contribute to the progress of technology and research.
Delve into my research endeavors on Google Scholar.
Experience
TikTok
Machine Learning Engineer
July 2024 - Present
E-Commerce Risk Control & Security - catching malicious buyers
Lucid Motors
Sr. Data Scientist
April 2024 - June 2024
Joined the team and quickly made remarkable contributions by leading the adoption of Generative AI for automating customer care data analysis. This initiative reduced manual workforce effort by 90%, streamlined operations, and provided valuable insights from customer feedback, resulting in potential significant process improvements.
Enabled the transition from rule-based to ML-driven anomaly detection for vehicle fleet security. This enhancement significantly reduced false positives by 50%, simplifying the validation of cybersecurity threats. Proposed and implemented feature importance techniques, which enhanced the explainability and reliability of vehicle security operations.
Nokia Bell Labs
Autonomous Systems Research Intern
June 2023 - August 2023
Leveraging large language models (LLMs) to enhance Nokia's patent-pending, proprietary MLOps platform for the end-to-end operations of ML-based use cases. Developing innovative task-specific knowledge enrichment strategies, involving automatic retrieval using Langchain and vectorstores, to improve the performance of LLMs in complicated code generation tasks. [Manuscript under review at IEEE Transactions on Artificial Intelligence]
SenticNet, NTU Singapore
Research Intern
May 2021 - July 2021
Developed a deep multitask learning framework that enhances the performance of Negation Scope Detection using POS tagging as an auxiliary task. Used transformers and neural tensor fusions to leverage the inter‑task correlations. Achieved 5% improvement over the baseline models. [ICDMW '21]
Mysuru Consulting Group
Machine Learning Intern
April 2020 - June 2020
Employed advanced Recurrent Neural Networks (RNNs), including Bidirectional Long Short-Term Memory networks (LSTMs), to forecast stock market excess returns, empowering clients with valuable insights and investment analysis.
Ernst & Young
Trainee with the Transaction Advisory Services team
December 2019 - January 2020
Designed a quantitative model for evaluating an industry’s attractiveness using Porter’s Five Forces Framework and SWOT Analysis.