Senior AI Engineer (LLM & Real-Time Systems)

About PubNubPubNub powers real-time experiences for 2,000+ companies including Verizon, Autodesk, Zillow, and Dropbox.Our global data network processes trillions of messages monthly with sub-100ms latency across 15+ data centers worldwide.We’re now building an AI capability layer that enables developers to add AI features (classification, summarization, routing, enrichment, automation) directly into real-time streams — without compromising latency, reliability, or trust.This is where you come in.What You’ll BuildYou’ll design and operate production AI services that integrate directly into PubNub’s real-time messaging platform.This is a systems + platform engineering role with applied AI, not research.You’ll work on:AI-powered moderation and enrichment pipelinesLow-latency inference systems running on high-throughput streamsInternal APIs, SDKs, and tooling that enable product teams to ship AI safelyObservability, evaluation, drift detection, and production debugging workflowsModel routing, retrieval patterns (RAG), batching, caching, fallbacksTrade-offs between latency, cost, accuracy, and privacyYou will not be training foundation models from scratch.Must Have5+ years backend / platform engineering experience1+ year shipping AI-enabled features in productionExperience integrating LLMs (OpenAI, Azure OpenAI, Bedrock, OSS models, etc.)Experience building high-throughput systems (streaming, queues, real-time APIs)Strong fundamentals in system design (performance, reliability, observability)Fluency in TypeScript, Python, or Rust(and willingness to work across ecosystems)Comfortable using AI-assisted development tools (Copilot, Cursor, Claude, etc.)Fluent EnglishNice to HaveReal-time systems (Kafka, Kinesis, WebSockets, pub/sub, event-driven design)Kubernetes / Docker / infra-as-codeModel serving tools (vLLM, Triton, TensorRT, TorchServe)Vector search / embeddings / RAG pipelinesExperience handling PII, compliance, safety guardrails in AI systemsWhy This Role Is InterestingYou’ll ship AI that runs in real time — not offline batch jobsYou’ll solve hard constraints: latency, scale, cost, trustYou’ll build internal platform primitives used across multiple teamsYou’ll work on greenfield AI systems with real production impactWhy PubNubRemote-first within PolandOptional office in central KatowiceCompetitive B2B compensation: 26 000 – 35 000 PLN net/monthWork on real production AI at global scaleEngineering-heavy culture (50%+ developers)If you want to build AI that works under real-world scale constraints, not just demos, we’d love to talk.