Welcome to Chenxi's personal page.

Profile

ML Engineer: 6+ years’ experience in AI related projects, hand on experiences on various AI models (regression, trees, NLP, LLM, NN), techniques (RAG, AI-agent, model fine tuning, and evaluation), end-to-end ML/AI project deployment (AWS EC2, Jenkins).
Scientist: 10+ years of physics-based model development and retrieval algorithm development. 10+ AI related Journal publications and conference papers. 5+ AI related NASA and NSF PI and Co-Is.
Front-end IOS Developer: published an IOS application in Apple Store
Full-stack Software Developer: 3+ years’ experience of full-stack frameworks using Python and NextJs. 1+ year experience of web-search engine development with Java, NextJs and Elastic Search.
NASA Student Internship Mentor: mentored 3 graduate students at UMD and UMBC and 4 NASA intern students in AI-related projects.
Skills: Python, TensorFlow, PyTorch, AWS, Elasticsearch, Java, PySpark, Scala, Databricks, JavaScript, Swift UI, SQL, Docker.

(2023 – 2024): Develop ML regression model for Legal Document ETL.
(2024 – current): Develop an AI-agent with open-source LLM (Llama-3), refined with domain knowledge to better understand users’ search queries in a search engine.
(2024 – current): Design and develop AI-chatbot system to support internal and external users.
(2024-current): Develop a backend search engine to support document retrieval (Java reactive).
(2022): Develop an IOS application using Swift UI.
(2018 – 2019): Train and apply AI-models to find global ship tracks from NASA satellite images.
(2020 – 2022): Develop a satellite training-dataset generating platform (python-based) for NASA researchers.
(2022 – 2023): Identify cloud/pollution/surface from Satellite Image Time-Series.
(2022 – 2023): Identify night-time wave-patterned air glows from noisy satellite images.
(2018-2023): Develop, improve, and maintain NASA cloud physics retrieval algorithm and data.
(2016 – 2018): Identify cloud/pollution/surface from multi-channel satellite data.
(2013-2015): Develop physics-based models and retrieval algorithms.
(2014-2016): Produce multiple satellite products for public users and researchers.

Apply AI models in various remote sensing applications.

Smart Search Engine (traditional Elastic Search + AI vector search system design)
AI assistant (Prompt Eng., RAG, LLM fine-tuning, AI-agents)
AI chatbot (Prompt Eng., RAG, LLM fine-tuning, AI-agents, Cache)