About this role
--- base: / meta-keywords: meta-viewport: width=device-width, initial-scale=1.0 title: System Engineer - AI Engineer (100%) --- System Engineer \- AI Engineer (100%) ===================================== We are looking for a System Engineer — AI Engineer to integrate open\-source models and the surrounding ecosystem into production services running on\-prem in our datacenters. You'll join an experienced and enthusiastic team with a great team spirit, where knowledge is shared rather than hoarded. The services are business\-critical, so the team shares a 24/7 on\-call rotation. System Engineer \- AI Engineer (100%) ===================================== **What you'll do** You own your solutions end to end — from understanding the business need to running them in production : * Analyse the need with stakeholders, challenge it, and turn it into a design that fits our constraints — then implement, test, document, and deliver. * Take ownership beyond the code: timelines, dependencies, communication with stakeholders, and the decisions that come with them. * Evaluate, select, and deploy open\-source models (LLMs, embeddings, rerankers, speech), and run the self\-hosted inference stacks behind them (vLLM, Ollama) with sane quantization, batching, and model routing. * Build the application layer — RAG pipelines, agents, tool calling, structured output — and own the data path behind it: ingestion, chunking, embeddings, vector stores, retrieval quality. * Build evaluation harnesses and benchmarks so model choices are measured, not guessed. * Integrate the ecosystem around the models: gateways, orchestration frameworks, model registries, observability and tracing. * Tune for latency, throughput, and GPU utilization; handle capacity planning, scheduling, and production monitoring — and debug regressions when a model or version changes. * Take part in the team's 24/7 on\-call rotation (piquet), with compensation and time off in lieu — and improve the runbooks, alerting, and automation so quiet evenings are the norm, not the exception. **What you bring** * Master's degree in Computer Science or equivalent, with hands\-on experience working with LLM\-based systems. * System engineering experience on Linux and Kubernetes. * Strong ; comfortable with production practices (testing, CI/CD, code review). * Strong SQL skills. * Experience self\-hosting and serving open\-source models. * Working knowledge of retrieval, embeddings, and vector databases. * Hands\-on: not afraid of cabling a server or swapping a GPU. * Autonomy and a sense of ownership — you follow a topic through to delivery rather than handing it over. * Willingness to join a 24/7 on\-call rotation, and the temperament to stay methodical under incident pressure. * Pragmatism about open\-source tooling — able to tell a solid project from a hyped one. Nice to have * GPU operations: drivers, CUDA, MIG, device plugins, scheduling on Kubernetes. * Inference optimization (TensorRT\-LLM, ONNX, speculative decoding, KV\-cache tuning). * Lightweight adaptation techniques (LoRA/QLoRA adapters on existing open\-source models). * Storage and networking design for AI workloads. * Experience with AI safety, guardrails, and responsible deployment practices. * Contributions to open\-source AI projects. **What we offer** * Friendly and dynamic environment where proactivity and personal commitment are highly valued. * Competitive salary and benefits system, very attractive pension fund conditions. * Flexible working hours with 6 weeks of holidays. * Modern and ergonomic workplace. * Various fringe benefits. * Free mobile subscription and discounts on our products. * Possibility to widen your skills and experience due to a fast\-moving and complex telco environment. * A full open\-source AI stack running on\-prem in our datacenters — you own the hardware, not someone else's abstraction. **Always dreamed of having a job in a diverse and refreshing environment? Then we really need to meet. Please apply online (we only accept online applications with CV, employment references, diplomas).**  * * *  * * * jid89aa0a3jm jit0940jm jiy26jm