Scientific data pipelines and infrastructure

– Built reusable statistical analysis workflows for large-scale time-series data and applied dimensionality reduction and sequence modeling to uncover latent states in complex system behaviors.

– Contributed to a multimodal deep-learning model for time-series classification by integrating high-dimensional features with temporal derivatives.

– Engineered a self-directed AI system that autonomously executes a live data pipeline with LLM-powered agents to provide real-time feedback and adaptive decision-making.

o Dockerized a web application for interactive neuronal recording data visualization using Plotly Dash framework and deployed to UCSC Genomics Institute public server

o Applied callback functions to link different graphs for information integration and user interaction

o Utilized AWS S3 database to store user-labeled neuronal data and text input metadata

o Developed algorithms to reveal insights from the data and further improved experiment design

o Shared functions and tools with collaborators and used Git for version control