Algorithm - Serving System Researcher
Tasks
- Build analytical performance models and roofline analysis
- Define next generation NPU serving research challenges
- Design scheduling batching and KV cache management algorithms
- Develop POC and prototypes with SW team collaboration
- Develop cost models and custom simulators
- Perform NPU based validation and implement prototypes
- Present research findings with technical documentation
Perks/Benefits
- N/A
Skills/Tech-stack
Analytical modeling | Attention | Batching | Cost modeling | Data Analysis | Data Modeling | Disaggregation | KV cache | Prefill Decode | PyTorch | Python | Python Data | Python data analysis | Roofline analysis | Scheduling | Simulation
Education
Roles
Related jobs
-
Audio Processing | Autoregressive models | Codec Models | Deep learning | Diffusion ModelsSenior-level Full TimeS.Korea-Seoul, Korea, Republic of9d ago
-
Audio Processing | Autoregressive models | Codec-Based Modeling | Deep learning | Diffusion ModelsSenior-level Full TimeS.Korea-Seoul, Korea, Republic of9d ago
-
Agentic reinforcement learning | Benchmark design | Deep learning | Experiment automation | JAXSenior-level Full TimeSeoul1mo ago