Yimei Wu

About Yimei Wu

My background bridges applied AI research and full-stack system development. I’ve worked on model fine-tuning, multimodal architectures, generative model, GPU-accelerated computation and deploying scalable ML infrastructure on the cloud. Highlights of my recent work include- Developed a modular video-to-music generation framework that aligns audio with visual transitions using CLIP-based semantic & temporal features, VAE latent encoding, and Transformer-based diffusion, with ongoing work on LLM-guided text conditioning for enhanced coherence- Fine-tuned GPT-2 with LoRA to enhance multi-hop reasoning in a medical RAG system- Deployed cloud-native microservices and CI/CD pipelines on GKE with automated analysis and Hadoop integration- Developed the Needle deep learning framework from scratch, enabling automatic differentiation and CUDA-based sparse-dense matrix multiplication operations to accelerate Graph Neural Network (GNN) training- Fine-tuned YOLOv8 for image classification and designed FGSM based adversarial attacks to test model robustness I’m passionate about building robust, real-world machine learning systems and excited to contribute to a mission-driven team.

Experience

1. Ai Research Engineer

   Archer

   Jul 2025 — Present