LLM算法/机器学习/AIGC算法（J106790）
Core
Research and implement core algorithms for AIGC advertising creative image generation, applying multi-modal large models, text-to-image/diffusion models, and reinforcement learning to improve generation quality, relevance, and commercial conversion.
Role type
Senior IC multi-modal large model and AIGC algorithm engineer
Builds
Industrial-grade Harness Agent workflows, Agentic AI architectures, and automated content moderation systems for ad creatives
Domain
Advertising technology, AIGC, multi-modal AI, reinforcement learning
Deliverable
production ML models
Required skills
Python, PyTorch, multi-modal large models, diffusion models, computer vision, NLP, reinforcement learning, SFT, DPO, RLHF, agent workflow planning, vector retrieval, multi-modal representation learning
Preferred skills
Publications in CVPR, ICCV, NeurIPS, ICLR, SIGGRAPH, strong mathematical foundation, ability to reproduce cutting-edge papers
Technologies
PyTorch, Harness Agent, SFT, DPO, RLHF, vector search
Responsibilities
Develop multi-modal large models and diffusion models for ad creative generation; research Agentic AI architectures and long-horizon task planning; build automated content moderation and quality filtering systems; optimize training stability and rollout efficiency for RL methods
Seniority
Senior, hands-on IC
