Algorithm Engineer – AIGC

Overview

This role focuses on designing and optimizing training pipelines for large-scale AI-generated content (AIGC) models. The engineer will handle data curation, ensure training stability, and optimize performance, specifically applying large-scale diffusion models to practical tasks like low-light image enhancement.

Key Responsibilities

• Develop and implement training pipelines for large-scale AIGC models emphasizing data preparation and model optimization.
• Design architectures and algorithms for AIGC and extensive multimodal models involving images and videos, covering both foundational training and subsequent alignment for quality improvements.
• Advance post-training alignment methods for diffusion-based multimodal models to enhance output quality.
• Explore and prototype new features such as multi-modal comprehension and visual content creation.
• Stay abreast of state-of-the-art AIGC research, manage project planning, deliver production-level solutions, and contribute to academia via publications in top-tier conferences.

Qualifications

• Master’s degree or higher in Computer Science or a closely related discipline.
• Publication as first author in premier scientific venues (e.g., CVPR, ICLR, ECCV, ICML, NeurIPS, TPAMI) focusing on diffusion-based multimodal generation and post-training alignment techniques that enhance generation quality.
• Practical experience with multimodal large models in image or video generation.
• Proficient in PyTorch for model development.
• Strong programming expertise in Python, C++, and CUDA.
• Excellent problem-solving capabilities, creativity, and effective teamwork and communication skills.