AI Platform Engineer

AI Platform EngineerQNB supports people eager to grow their careers and shape the future of the sector, in an equal, fulfilling and empowering environment.We're growing our AI organization and looking for an engineer who knows what it takes to run AI systems in production and to build the platforms behind them. Working side by side with our AI Engineers, you'll do two things: take the agents they design to production — reliable, observable, secure and business-critical on our own GPU infrastructure — and build the platforms that let technical and non-technical users create, test and manage agents. Everything is built in-house, from the platform up.In this role, you willTake agents to productionTake LLM-based agents from prototype to production: deployment, scaling, versioning, autoscaling, zero-downtime rollout and rollback, disaster recovery.Deploy and optimize self-hosted LLM inference (e.g. vLLM) on GPU infrastructure on OpenShift/Kubernetes — capacity planning, GPU utilization, batching, quantization, multi-model and LoRA serving.Load, stress and soak test agentic systems end to end — LLMs, MCP servers, vector databases and internal APIs — measuring throughput, TTFT and inter-token latency, finding the bottleneck and turning results into capacity plans.Build observability for AI workloads: structured logging, metrics, distributed tracing and LLM-specific telemetry such as token usage, cost and agent/tool-call traces.Write resilient applications — timeouts, retries with backoff, circuit breakers, graceful degradation — and apply security and compliance for a regulated, on-prem environment: authentication for agents and MCP servers, secrets management, network isolation, audit logging and PII handling.Build and run the agent platformBuild platforms for the full agent lifecycle: agent builders, simulation and testing environments, MCP server and connector registries, prompt and version management, deployment and monitoring.Build the agent runtime at the core: orchestration loop, tool calling and MCP integration, context, memory and state management, guardrails, and sandboxed execution of untrusted code.Make platforms work for everyone: low-code interfaces for business teams, SDKs, APIs and CLIs for engineers.Own them long-term: gather feedback, shape the roadmap, drive adoption, train users, and build golden paths so AI Engineers ship safely without reinventing infrastructure.We are waiting for your application if you haveA bachelor's degree in Computer Engineering, Software Engineering or a related fieldMinimum 4 years in software, backend, platform, SRE or ML infrastructure engineering, having taken high-criticality, 24/7 services live and kept them healthyStrong Python and solid engineering practices (testing, code review, clean architecture); Go is a plusKubernetes and/or OpenShift: deployments, Helm/Operators, resource management and GPU scheduling (e.g. NVIDIA GPU Operator, MIG)Experience serving LLMs with vLLM or similar engines (TGI, SGLang, Triton, KServe)Hands-on performance testing (k6, Locust), LLM benchmarking (GuideLLM, LLMPerf) and observability tooling (Prometheus, Grafana, OpenTelemetry, ELK/Loki, Jaeger/Tempo)REST/gRPC APIs, async programming, microservices, Kafka; SQL with PostgreSQL and RedisWorking knowledge of agent frameworks (LangChain, LangGraph), RAG, vector databases and MCP — enough to productionize, load test and troubleshoot themAwareness of security, data privacy and compliance in regulated environments; fluent EnglishAn ownership mindset ("you build it, you run it"), product thinking, and a passion for building platforms people actually useIt's a plus if you haveLow-code agent platforms (Langflow, Dify, n8n) or frontend experience (React) for platform interfacesAI workloads on-prem or air-gapped; untrusted code in sandboxes; LLM evaluation tools (Langfuse, Arize Phoenix)Real-time speech (STT/TTS) under latency constraints; GitOps (ArgoCD, Tekton); chaos engineering; GPU capacity planning; fine-tuning pipelines (LoRA) through to servingIf you know how to make AI systems run reliably at scale and want to build the platforms behind a large bank's AI efforts, we'd like to hear from you.