Mark C.

Since 2024.09 I'm in charge of the Tiktok AI Search, which serves hundreds of millions of American users everyday. We aim to use LLM techniques to improve the experience of users. I worked on everything related to AI Search. ---- I work on LLM foundation models and code generation models. I have been working on pretraining, finetuning and evaluation of the models. Our product --- Doubao, has more than 30 million MAU right now. Our code generation model is on par with the SOTA opensource code models.

Fusion: Design and implement multi-sensor fusion models, algorithms, and pipelines. Using techniques like DeepLearning, XGboost, Graph Neural Network and etc via ML frameworks such as MXnet and Pytorch. Deploy algorithms and models onto our trucks using c++ with GPU support. Tracking Design and implement multi-frame multi-object tracking models,algorithms and pipelines. Using techniques like DeepLearning, XGboost, GCN, RNN and kalman filters Machine Learning Platform/Pipeline: Developed the first auto sampling, auto training, and auto benchmarking platform for perception modules—processing Terabytes of road test datasets daily and updating our models when deemed necessary by the algorithm.

1. Designed and built a git-style data management system that provides data versioning and can be used both in the command line and raw python project. 2. Used recursive hashing to compare folder difference and enabled the system to track terabytes size datasets. 3. Made a central data sharing system that can push and pull(download) datasets using a single command. Greatly reduce the time needed for sharing datasets (instantly for pull). 4.Reduced 20~30% of disk usage on EFS by providing symlinks/refinks to users and save $12~20K per month for the team potentially. 5. Help to build one of Amazon Search's first Bert based relevance model

Master of Science in Computer Science (MSCS) student Honored to study with some of the best minds in the world

Honor Student Scholarship for Top 10, 2015 & 2016; Minor in Business and Management, GPA 91/100 Rank 1/256 in Principles of Economics course;