Research Scientist – Model Efficiency

About the role

Bitdeer AI Lab is seeking a Research Scientist focused on model efficiency to make AI models cheaper and faster to serve without sacrificing quality. You will work on cutting‑edge techniques that improve inference performance for large language models across our proprietary hardware and datacenter infrastructure.

Key responsibilities

• Implement and adapt published efficiency methods such as quantization, sparsity, pruning, speculative decoding, MTP, serving‑time attention, and KV‑cache.
• Develop original optimizations when existing methods fall short.
• Build rigorous evaluation pipelines to validate claims like “lossless at 2× throughput”.
• Collaborate with engineering teams to bring efficiency techniques into production serving environments.

Required profile

• Bachelor's, Master's or PhD in Computer Science, Electrical Engineering, or a related field.
• Hands‑on experience with LLM inference, model optimization, or ML systems.
• Strong programming ability in Python and deep familiarity with PyTorch; experience with C++, CUDA, or Triton is a plus.
• Demonstrated depth in at least one model‑efficiency area (e.g., quantization, sparsity, speculative decoding).
• Solid understanding of transformer internals and where accuracy loss occurs.
• Rigorous evaluation practice with honest baseline comparisons.

Required skills

• Python
• PyTorch
• C++ (optional)
• CUDA (optional)
• Triton (optional)
• Model quantization
• Sparsity and pruning
• Speculative decoding
• MTP
• Serving‑time attention
• KV‑cache management
• Transformer internals
• LLM inference