Sirius Chen

LLM Post-Training (2024-2025) -Core contributor for Llama post-training and evaluation; unblocked critical infrastructure and model pipelines, enabling final model launch with top-tier benchmarks. -Optimized evaluation stack and inference pipeline (2–5× speedup). -Pioneered hybrid reasoning training recipes and evaluation frameworks. Multimodal LLMs (2023-2024) -Led design and deployment of MetaAI joint video–image LLM, including training recipe, evaluation, and production rollout. -Built multimodal LLMs for content understanding and partnered cross-functionally to integrate models into FB Search, Feed Ranking, and IG Reels, driving measurable engagement improvements. Visual Understanding (2020-2023) -Developed models for misinformation and sensitive content detection, deployed at production scale. -Built image/video embedding models that improved recommendation quality

Research: Conduct computer vision research with professor Larry Davis Courses: CMSC 660 Scientific Computing by professor Ramani Duraiswami CMSC 828G Image Understanding by professor Ramalingam Chellappa ENEE 698A Graduate Seminar (topic: deep learning) by professor Ramalingam Chellappa Teaching Assistant: CMSC 122 Introduction to Computer Programming via the Web by professor Fawzi Mead

Developed a algorithm for cross-age face recognition. Paper accepted at ECCV 2014.

Mandatory military service. I was responsible to maintain a wired communication system.

Internship with professor Shin'ichi Satoh on Large-Scale Face Image and Video Retrieval

Teaching assistant to professor Shou-De Lin: Probability, Spring 2011 Teaching assistant to professor Winston Hsu: Database System, Fall 2010

High school exchange student Senior year in Olpe High School, Kansas, USA