腾讯专有云-大模型推理优化工程师-深圳/北京
Core
Designing and optimizing inference architecture and engines for private cloud MaaS platforms to serve enterprise customers.
Role type
Senior IC large model inference optimization engineer
Builds
Private cloud MaaS platform inference capabilities for enterprise deployments
Domain
Cloud infrastructure + Large Language Models (LLMs)
Deliverable
production ML models
Required skills
Large model inference optimization, High-performance computing, Inference engine customization, Heterogeneous hardware adaptation, Distributed inference (tensor/pipe parallelism), Quantization, Continuous batching, KV Cache management, Attention variants, System profiling & benchmarking
Preferred skills
C++ / CUDA / Python, vLLM / SGLang / TensorRT-LLM / MindIE / lmdeploy / TGI, Speculative decoding
Technologies
vLLM, SGLang, TensorRT-LLM, MindIE, lmdeploy, TGI, CUDA, Python, C++
Responsibilities
Design inference architecture (scheduling, routing, multi-instance/multi-node deployment), Optimize inference engines for throughput/latency/memory, Adapt open-source models to private cloud, Integrate AI accelerators, Develop reusable acceleration solutions, Build performance baselines and regression systems
Seniority
Senior, hands-on IC
