top of page
me_full_edited.jpg

welcome

I’m a machine learning researcher interested in the intersection of Human and Artificial Intelligence.

M.S. in Computer Science, 2023

University of Massachusetts, Amherst

B.S. in Computational Mathematics, 2020

University of California, Los Angles (UCLA)

Research Experience

Jan '23 - May '23

Unified NLU Reranker with Deep RL

Natural Language Understanding systems like Alexa use several language processing models that generate possible hypotheses for user commands. An agent then ranks them, accounting for calibration errors in models. However, most datasets and models are proprietary. We addressed this by coding a sandbox for re-ranker agents using a synthetic data generation algorithm.

Advisor: Dr. Yuguang Yue

Jun '22 - Dec '22

Representation Learning in Hierarchical RL

Probed existing HRL methods for challenges and limitations. I also designed a novel algorithm that addresses those challenges to provide more robust and efficient learning through extracting smart representations of the environment.

Advisor: Dr. Bruno Castro da Silva

Jan '22 - May '22

Image Dense Captioning for Child Rescue

Prototyped and tested an image captioning DenseCap ML model to extract objects and captions of relevance from doorbell cameras and allow canvassers to query them for accelerated and guided search through videos.

Projects

Jan '22 - May '22

Designed a novel algorithm that uses Behavior Cloning to learn from expert demonstrations and calls experts in an online setting only when needed, increasing data efficiency by a factor of 4‑6 while maintaining final performance.

Apr '20 - Jun '20

Stock Modeling using Relational Learning

Adapted a Graph Convolution‑LSTM model to PyTorch using Deep Graph library, leveraging embeddings from industry relations from WikiData and sector‑industry datasets, and successfully replicated optimal results on stock price data.

Apr '18

LA Hackathon: Fear Me videogame

Developed a proof‑of‑concept video game was developed at the annual UCLA Hackathon. The idea, implemented in code, was to change the fear level of a video game based on how scared the player was: if the game gets too scary, the game would reduce its fear factors. Won the Best Gaming Hack Award.

bottom of page