Research Engineer-Model Efficiency
Core
Optimize large language model inference to reduce cost and latency while maintaining quality through implementation of quantization, sparsity, pruning, speculative decoding, and serving-time attention methods.
Role type
Research Engineer specializing in model efficiency and LLM inference optimization
Builds
Optimized inference pipelines and evaluation frameworks for Bitdeer's AI cloud and HPC datacenters
Domain
Artificial Intelligence / Large Language Models / High-Performance Computing
Deliverable
production ML models
Required skills
LLM inference optimization, model quantization, sparsity and pruning, speculative decoding, transformer internals, rigorous evaluation practice, Python, PyTorch
Preferred skills
C++, CUDA, Triton, inference engines (vLLM, SGLang, TensorRT-LLM), production serving experience, top-tier publications
Technologies
PyTorch, C++, CUDA, Triton, vLLM, SGLang, TensorRT-LLM
Responsibilities
Implement and adapt published model efficiency methods on company models and hardware, develop custom optimizations where existing methods fall short, build evaluation discipline to validate performance claims, take efficiency methods into production serving
Seniority
Mid-Senior, hands-on IC