Senior Capability Development Engineer

NVIDIA NVIDIA · Semiconductors · Shenzhen, China

NVIDIA is seeking a Senior Capability Development Engineer to develop and enhance internal RAG and Agent platforms for Ops Engineering productivity. The role involves developing, training, fine-tuning, and deploying multimodal LLMs, building LLM-based applications (RAG, TEXT2SQL, Agents), applying advanced tuning techniques, measuring performance, analyzing accuracy/bias, and driving dataset development. Requires strong Python skills, familiarity with ML/DL frameworks and LLMs, and practical experience with LLM training frameworks.

What you'd actually do

  1. Develop, Train, Fine-tune, and Deploy (multimodal) large language models for various AI-based applications.
  2. Develop LLM-based applications for such as RAG, TEXT2SQL, Agents.
  3. Apply instruction tuning, reinforcement learning with verifiable rewards (RLVR), and parameter efficient fine-tuning such as LoRA to improve LLMs for various AI-based applications.
  4. Measure and benchmark model and application performance.
  5. Drive the gathering, building, and annotation of domain specific datasets to train LLMs for different tasks and applications

Skills

Required

  • Python
  • ML/DL algorithms and frameworks
  • LLMs
  • Conversational AI technologies
  • RAG
  • Text2SQL
  • Agents
  • Embeddings
  • Rerankers
  • TRL
  • VERL
  • OpenRLHF

Nice to have

  • AI coding assistants (e.g., Cursor, Copilot)
  • Systems Thinking
  • dataset cleaning
  • Data Mining
  • RLVR/Continue Pre-training/SFT
  • SW architecture designing

What the JD emphasized

  • over 8 years of relevant work experience
  • Strong programming skills in Python
  • Deep understanding of ML/DL algorithms and frameworks
  • Practical experience training LLMs using frameworks such as TRL, VERL, or OpenRLHF

Other signals

  • Develop LLM-based applications
  • Enhance NVIDIA internal RAG and Agent platforms
  • Develop, Train, Fine-tune, and Deploy (multimodal) large language models