Эта вакансия в архиве
Посмотреть похожие вакансии ↓3 месяца назад
Research Scientist-Model Efficiency (AI)
Описание вакансии
Текст:
TL;DR
Research Scientist-Model Efficiency (AI): Making large language models cheaper and faster to serve without sacrificing meaningful quality, with an accent on quantization, sparsity, speculative decoding, and serving-time attention and KV-cache optimization. Focus on implementing and adapting published methods, developing new inference optimizations, and designing rigorous evaluations that validate throughput and quality claims.
Location: Singapore, SG or Austin, US
Company
is a technology company focused on Bitcoin mining, AI cloud services, computing infrastructure, and high-performance computing datacenters.
What you will do
- Optimize large language models for faster and more cost-efficient production inference.
- Implement and adapt methods including quantization, sparsity, pruning, speculative decoding, MTP, and attention or KV-cache optimization.
- Develop new model-efficiency techniques when existing approaches do not meet quality or performance requirements.
- Adapt optimizations to internal models and hardware platforms.
- Build rigorous evaluation practices using task-level metrics, controlled comparisons, and honest baselines.
- Collaborate on production serving and AI infrastructure improvements.
Requirements
- Bachelor’s, Master’s, or PhD in Computer Science, Electrical Engineering, or a related field.
- Hands-on experience with LLM inference, model optimization, or machine learning systems.
- Strong programming skills in Python and deep familiarity with PyTorch.
- Implementation-level expertise in at least one model-efficiency area, such as quantization, sparsity and pruning, speculative decoding, MTP, or serving-time attention and KV-cache methods.
- Strong understanding of transformer internals, accuracy loss, and model behavior, together with rigorous evaluation practice.
- Experience with production inference, inference engines such as vLLM, SGLang, or TensorRT-LLM, and technologies such as C++, CUDA, or Triton is advantageous or highly preferred depending on the area.
Nice to have
- Publications at top-tier systems or machine learning venues.
- Substantial open-source contributions.
- Experience taking efficiency methods into production serving.
Culture & Benefits
- Inclusive environment that values authenticity and diverse perspectives.
- Startup spirit within a fast-growing technology company.
- Opportunities to contribute directly to new AI infrastructure and digital asset projects.
- Autonomy, personal accountability, and opportunities for rapid growth and learning.
- Training, mentoring, welfare benefits, and professional development opportunities.
Музыка
Осень
Поддержка