Director, Model Inference Engineering
thomsonreuters.wd5.myworkdayjobs.com
- Job written in
- English
- Location
- Zug
- Work type
- Hybrid
- Type
- Full-time
Thomson Reuters Labs, the research and development arm of Thomson Reuters, creates AI‑driven products for high‑stakes professional workflows. The team is building Mariner, a custom inference platform that serves the company's proprietary LLM family and benchmark models across multiple products and regions. As a Director‑level, hands‑on engineer you will help shape and run this serving stack, ensuring it scales reliably while meeting cost, latency and compliance goals. Your day‑to‑day work will involve designing, building and operating the model‑serving infrastructure that handles all production traffic. You will implement inference‑optimisation techniques such as batching, caching and quantisation, improve observability as call volume grows, and establish ISO 27001‑related controls, cost‑tracking mechanisms and GPU‑capacity contingency plans. You will also document trade‑offs between cost, latency and reliability, collaborate with training and evaluation pods to launch new model releases, and set up team practices like on‑call rotation, incident reviews and onboarding. The role requires solid hands‑on infrastructure experience, especially with Kubernetes or a comparable orchestration system, and a proven ability to own production services end‑to‑end. You must have worked on model‑serving or inference optimisation for large models, managed large‑scale distributed systems under peak load, and demonstrated strong debugging skills for production incidents. A track record of shipping software from design through operation, enforcing engineering best practices, and mentoring junior engineers is also essential. Preferred experience includes familiarity with common LLM‑serving stacks such as vLLM, TensorRT‑LLM, Triton or SGLang, exposure to GPU‑cluster capacity planning or multi‑region deployments, and work in regulated environments (ISO 27001, SOC 2). Knowledge of systems languages like Rust or Go and expertise in low‑level optimisation of rate‑limiting, routing and hardware bottlenecks are considered a plus. The position offers a hybrid work model, flexible "Flex My Way" policies, up to eight weeks of remote work per year, extensive benefits, and a culture focused on continuous learning and inclusion. What the role asks for: - Hands‑on infrastructure engineering with Kubernetes or similar - Experience operating model‑serving infrastructure at scale - Large‑scale distributed systems design and reliability expertise - Proven end‑to‑end software shipping track record - Ability to mentor and grow team members - Experience with LLM‑serving stacks (vLLM, TensorRT‑LLM, Triton, SGLang), nice to have - Exposure to GPU‑capacity planning or multi‑region deployment, nice to have - Familiarity with ISO 27001 or SOC 2 compliance environments, nice to have
Engineering: Lebenslauf-Vorlage
Bewirb dich mit einem Lebenslauf im Schweizer Aufbau — mit Beispieltext und den Anforderungen, die Engineering-Inserate am häufigsten nennen.
Lebenslauf-Vorlage Schweiz ansehenEngineering: was der Markt gerade verlangt
Engineering-Stellen gehören zu den regelmässig ausgeschriebenen Berufen auf SwissJobs.app.
Weitersuchen
- Alle Geschäftsführer Jobs in der Schweiz
- Alle Jobs in Zug
- Lebenslauf als Engineering: Beispiel und Vorlage
- Lebenslauf mit KI erstellen
- ATS-Lebenslauf prüfen
- Bewerbungsschreiben für dieses Inserat