Machine Learning @ TikTok Search | CS + Financial Engineering @ WashU
building AI systems people actually use.
LLM post-training, AI agents, retrieval and ranking, medical AI, and teaching, with industry work closest to real users.
Industry
My industry work has put me close to practical relevance problems: ranking quality at TikTok Search, plus LLM post-training, model quality, and evaluation loops at Meituan.
Machine Learning Engineer Intern, TikTok Search
Working on the fine ranking core algorithm for TikTok Search, where user intent, relevance quality, and large-scale ranking infrastructure meet.
Machine Learning Engineer Intern, Search and Content Intelligence
Focused on LLM post-training and enterprise AI quality loops, with model compression, vLLM inference, compliant PII handling, and retrieval evaluation around Meituan-scale products (770M+ users and 15M+ merchants).
Research
My research sits where agents meet accountability: medical image segmentation, grounded clinical reasoning, retrieval, and restoration under hard evaluation.
Test Dice from validation-gated, safety-aware self-improvement for tumor segmentation.
Clinical QA accuracy on ACR TXIT with citation-grounded answer synthesis.
Zero-shot restoration gains under severe Gaussian noise and compound degradation.
Teaching
I have supported algorithms, cloud computing, data mining, ML, and statistics courses through recitations, office hours, grading systems, exam support, rubric calibration, and TA mentorship.
Head TA across 5 semesters; each semester, I coordinated 30+ TAs, supported 200+ students, and led recitations that made proofs, runtime analysis, and implementation patterns feel usable.
Head TA for 1 semester; helped students reason through cloud projects, infrastructure debugging, and system tradeoffs.
Clarified probability and statistics through office hours, exam support, and rubric calibration.
Guided applied data mining work, ML assignments, and grading consistency.
Supported model reasoning, programming assignments, and debugging sessions.
Led Data Structures and Algorithms practice from proof ideas to implementation instincts.
Honors + Skills
This section is intentionally compact: enough signal for context, without turning the site into a transcript.
One junior selected across all six Computer Science & Engineering majors at WashU for academic excellence.
Four-year merit scholarship at Washington University in St. Louis.
Recognized every semester enrolled.
S.-T. Yau CS Gold Prize, MCM Honorable Mention, Physics Unlimited 11th in the U.S.
Impact Board
High-impact algorithm work inside one of the world's largest recommendation and search ecosystems.
Post-training, model quality, PII-safe enterprise use, and retrieval evaluation.
Latency reduction through inference and CUDA profiling work.
Test Dice from a validation-gated self-improving agent framework.
Accuracy on a radiation oncology benchmark with grounded retrieval.
Training speedup for a large-action-space board-game agent.
Acc@0.5 on RefCOCO-M validation queries.
Ray, KubeRay, FastAPI, and artifact pipelines for distributed ML training.
Off Hours
I write lyrics, cook, backpack through Death Valley, and gravitate toward moody, poetic storytelling like Yorushika. Often with my cat nearby.
Contact
Talk to me about research, internships, projects, classes, or any topics you are excited about. I am always happy to meet and build!