ByteDance - Douyin Ads

May 15, 2026 · 1 min read
internships

Overview

Model Engineering Intern May 2026 – Present

Recommendation Model Migration

  • Reconstructed and upgraded ByteDance’s internal RL-based retrieval recommendation models from TensorFlow to LGTorch, including dual-tower architectures, vector quantization (VQ), and multi-runstep graph execution pipelines.
  • Designed sequence encoding tower decomposition, weight normalization alignment strategies, and write-back mechanisms for frequency estimation modules.
  • Completed full-stack migration across sparse feature processing, dense computation, and online serving pipelines.

Precision Alignment Infrastructure

  • Built a bidirectional forward/backward diff toolchain for end-to-end TensorFlow–LGTorch numerical verification.
  • Resolved feature-slot registration mismatches, VQ codebook numerical deviations, and framework-specific execution inconsistencies.
  • Achieved operator-level precision alignment between both frameworks; the associated technical documentation was classified as internal L3 confidential.

Production Model Upgrade

  • Upgraded Douyin’s RL-based retrieval pipeline with logQ debiasing and improved sampled-softmax estimation over sharded candidate repositories.
  • Achieved online performance comparable to the production baseline and successfully supported full-scale deployment.
Recommendation Systems Reinforcement Learning Ads LGTorch TensorFlow
YiFan Cai
Authors
YiFan Cai (he/him)
Graduate student
I am a graduate student in Systems Engineering at the University of Pennsylvania, with a B.S. in Computer Science from ShanghaiTech University. I have research experience in diffusion models, computer vision, robotic manipulation, and molecular drug design. My current interests focus on world models, computer vision, and artificial intelligence.