Senior AI Compute Infrastructure Engineer
Kraken · Remote · United Kingdom · Brazil · Ireland · Peru · Costa Rica · Latvia · Argentina · Romania · Cyprus · Portugal · Slovenia · Lithuania · South Africa · Czech Republic · Panama · Estonia · Canada · Hungary · Mexico · Poland · Spain · Bulgaria ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Senior level role
- Fully remote
- GPU & Kubernetes
- Python proficiency
- AI infrastructure
Join the AI Compute and Infrastructure team at Payward, the parent company of Kraken. You will own the infrastructure layer that enables fast, reliable, and cost-efficient AI workloads, including GPU cluster operations and model serving.
This position focuses on building production-grade systems for training, evaluation, and experimentation. You will collaborate with researchers and engineers to remove bottlenecks and ensure high availability for critical AI services.
What you'll do
- Operate GPU and accelerator clusters for training and inference workloads
- Design infrastructure to run models locally and reduce external dependencies
- Improve scheduling, orchestration, and quota management systems
- Optimize inference pipelines for latency, throughput, and cost efficiency
- Build observability tools for GPU utilization and system performance
- Drive reliability through incident response and post-incident improvements
- Evaluate new hardware and cloud instances for evolving AI needs
- Create tooling to simplify GPU usage for internal teams
What you'll need
- 5+ years of infrastructure engineering experience
- Hands-on experience operating GPU clusters in production
- Strong fundamentals in Linux, networking, and Kubernetes
- Experience with ML serving frameworks like vLLM or Triton
- Proficiency in Python for automation and tooling
- Understanding of performance tradeoffs in GPU computing
- Track record of optimizing compute costs and reliability
- Experience building observable systems with metrics and alerts
Nice to have
- Experience at an AI lab, hyperscaler, or HFT firm
- Familiarity with custom silicon like TPUs or AWS Trainium
- Background in capacity planning and GPU fleet cost management
- Experience with distributed training frameworks like DeepSpeed
- Debugging skills for CUDA, NCCL, or low-level performance issues
- Knowledge of Rust, C++, Go, or other systems languages
- Experience in crypto or financial services infrastructure
Stack
- Python
- Kubernetes
- Linux
- vLLM
- Triton Inference Server
- TensorRT
- Rust
- C++
- Go
- CUDA
How you'll work
- Fully remote position
- Global candidate eligibility
- High-stakes, always-on environment
Hiring process
- Application review
- Job-related skills or work-style assessments
- Interviews with team members
- Ongoing application acceptance