ML infrastructure and backend services for remote-first tech teams
Pragmatike is a Paris-based services firm completing ML and backend infrastructure projects for startups and scaleups. The stack reveals deep GPU/CUDA specialization (PyTorch, vLLM, Llama-factory, Megatron-LM, NCCL, NVLink) alongside cloud infrastructure (GCP, AWS, Azure, Kubernetes). They're actively migrating backend services from Node.js and Rails to Go while adopting CUDA and GPU-optimized frameworks — a pattern indicating focus on performance-critical, computationally intensive workloads rather than traditional web services.
Notable leadership hires: Tech Lead, Machine Learning Director, Head of Machine Learning
Pragmatike builds infrastructure and machine-learning systems for remote-first companies, focusing on deployment pipelines, GPU optimization, and multi-cloud orchestration. The business model centers on completing discrete technical projects — model serving infrastructure, Kubernetes clusters, peer-to-peer communication systems, and backend service migrations. The hiring velocity is accelerating with 468 roles posted in the last 30 days, heavily concentrated in engineering (610 open positions across seniority levels). They operate across 24+ countries including Canada, Brazil, Eastern Europe, and the US, supporting a distributed workforce model aligned with their customer base.
ML infrastructure, GPU optimization, backend services, and Kubernetes deployments. Active projects include building key management systems, designing production ML systems end-to-end, custom CUDA kernel development, and multi-cloud cluster orchestration.
Across 24+ countries including Canada, Brazil, Colombia, US, France, and much of Eastern Europe (Poland, Ukraine, Romania, Estonia, Lithuania, Latvia, Czechia, Hungary). Remote-first hiring model.
Other companies in the same industry, closest in size