Research Scientist-Model Efficiency (Intern)
Bitdeer Technologies Group is a technology company providing Bitcoin mining solutions, mining hardware, data-center infrastructure, and AI cloud services. It serves individual, institutional, and enterprise customers globally.
Funding history
Investors
Projects
About Bitdeer Technologies Group
Bitdeer provides vertically integrated Bitcoin mining and high-performance computing services. Its operations include mining equipment procurement and manufacturing, datacenter design and construction, equipment management, daily mining operations, cloud mining, and mining-related services. The company also offers AI cloud infrastructure and high-performance computing powered by NVIDIA GPUs for AI and machine-learning workloads, serving customers across global markets.
Skills
About the Role
Complete and present a well-defined three to six month project focused on making AI models cheaper and faster to serve. Implement or adapt model efficiency methods, develop optimizations, and use rigorous evaluation to produce a defensible result.
Requirements
- Undergraduate, Master's, or PhD candidacy in Computer Science, Electrical Engineering, Mathematics, or a related field
- Ability to commit 3–6 months
- Python programming ability
- Hands-on familiarity with PyTorch
- Coursework, self-study, or research experience in model efficiency
- Understanding of transformer internals
- Experience with rigorous evaluation
- Depth in a project, paper, or serious open-source contribution
Responsibilities
- Complete a well-defined model efficiency project
- Implement published model efficiency methods
- Adapt methods to models and hardware
- Develop model optimizations
- Evaluate task-level metrics and controlled comparisons
- Present the project results
Benefits
- Welfare benefits
- Training and mentoring
- Networking with industrial pioneers and enthusiasts
- Personal accountability, autonomy, fast growth, and learning opportunities
- Opportunity to contribute directly to digital asset industry projects