Generative AI Algorithms Engineer
Core
Research and develop algorithms for on-device generative AI models (LLM, LVM, VLM, omni) to enable state-of-the-art networks to run on devices with limited power, memory, and computation.
Role type
Senior IC algorithms engineer (generative AI)
Builds
Advanced AI technology and user-friendly model optimization tools (e.g., Qualcomm Innovation Center's AI Model Efficiency Toolkit) deployed in industry-leading devices.
Domain
Hardware/Software Systems Engineering + Generative AI
Deliverable
production ML models
Required skills
LLM/LVM/VLM/Omni model architecture design, MoE (expert routing, load balancing, sparse activation), quantization algorithms (FlexRound, PrefixQuant, GPTQ, SpinQuant, AMP), training/fine-tuning (pre-training, SFT, PEFT, RL, loss-function design), efficient inference (batching, KV caching, speculative decoding), model compression, neural search, Python, PyTorch
Preferred skills
Hardware Engineering, Software Engineering, Systems Engineering
Technologies
PyTorch, Python, FlexRound, PrefixQuant, GPTQ, SpinQuant, MoE
Responsibilities
Research and develop algorithms for LLM, LVM, VLM, and omni models; Develop advanced architectures for complex generative models including MoE; Develop quantization algorithms for complex generative models; Conduct experiments on generative models training and fine-tuning; Develop efficient inference algorithms; Perform model compression and neural search; Apply solutions for model efficiency advancement on device and in the cloud.
Seniority
Senior, hands-on IC
