CV
Download CV
EDUCATION
- University of California, Los Angeles (Los Angeles, CA)
- Master of Electrical Computer Engineering (Anticipated July 2027)
- University of Rochester (Rochester, NY)
- B.S. in Computer Science, B.S. in Business, Cluster in Studio Art (May 2025)
- Honors: Highest distinction, cum laude, Dean’s list (8/8 semesters), Schwartz Discover Research Grant 2022
ENGINEERING EXPERIENCE
- Student Researcher, Seizure Lab, ECE Department, University of California, Los Angeles
- Aug 2025 - Oct 2025
- Scaled VLM inference: deployed 3 models, boosted throughput 16.7× (0.3→5 rps) via asynchronous request batching, optimized latency by introducing a caching layer using prefix-cache hit ~73%.
- Built a medical visual dataset of 10k records on 8 features on epilepsy seizure. (CVPR 2026 under review)
- Student Researcher, Professor Chenliang Xu’s Lab, Computer Science Department, University of Rochester
- May 2025 - Sep 2025
- Conducted research on VLM-guided audio remixing’s improvement; transferred experience from traditional movie composers. Achieved SOTA with ~9% fewer parameters using a simplified architecture and prompt-variant sweeps; First author, submitted at IEEE ICASSP 2026.
- Built an async, sharded caption-embedding pipeline; boosted throughput 16× and raised KV-cache hits 6%→57% via chunking + prefix-cache; prototyped gated fusion & multi-head attention Module.
- Student Researcher, Professor Jiebo Luo’s Lab, Computer Science Department, University of Rochester
- Aug 2024 - Apr 2025
- Reconstructed animal images from skeletons using Stable Diffusion (I2I); applied partial diffusion to retain segmentation and sparse-cloud encoders to inject spatial cues—bridging long domain gaps.
- Improved facial realism via RLHF (DPO) with LoRA fine-tuning; built preference pairs and tuned prompts/schedules on UR Linux clusters.
- Machine Learning Internship, Farsee2 (Wuhan, China)
- July 2024 - Sep 2024
- Cut floating artifacts ~40% in Gaussian Splatting by adding angle-of-incidence and depth/normal supervision losses for better geometric consistency.
- Built customizable generation scripts with multi-machine parallelism; integrated into the product pipeline to improve throughput and reliability for prototype demo.
- Student Researcher, Goergen Institute for Data Science, University of Rochester
- Jan 2024 - Jun 2024
- Designed two ideology-bias metrics and a blind-test harness; automated data collection and reproducible scoring for LLM political-ideology evaluation.
- Found LLMs rate manifestos ~73.64% less extreme than ground truth; paper accepted in Journal of Political Institutions & Political Economy (SSRN: https://ssrn.com/abstract=4907043).
PROJECTS
- Body Signals Analysis of Smoking and Drinking
- Sep 2023 - Dec 2023
- Preprocessed over 900k records by selecting relevant attributes through feature engineering; applied PCA to 2 principal + 2 auxiliary components, capturing the dominant variance for compact modeling.
- Trained SVM and XGBoost for 5 categories of smoking/drinking prediction with 72/81% accuracy.
RESEARCH PAPER AND PREPRINTS
- Caliskan, Cantay and Huang, Junhua and Huang, Yiyang and Lin, Ruoxuan and Shan, Wanting, Using Generative AI to Calculate Party Positions: A Comparison of Human Experts and Large Language Models (July 26, 2024). Available at SSRN: https://ssrn.com/abstract=4907043
- Pinxin Liu, Luchuan Song*, Junhua Huang, Chenliang Xu, GestureLSM: Latent Shortcut based Co-Speech Gesture Generation with Spatial-Temporal Modeling. ICCV 2025, https://arxiv.org/abs/2501.18898
- Junhua Huang*, Chao Huang, Chenliang Xu, Semantic Visually-Guided Acoustic Highlighting with Large Vision-Language Models. Submitted to ICASSP 2026 (Sep 17, 2025).
