We’re supporting a fast-growing enterprise building an internal LLM platform for multiple business units (Arabic/English). They need an AI Platform Engineer to turn prototypes into production features—secure, observable, and cost-controlled.
Own the path from POC → prod for LLM features (chat, agents, document QA). You’ll standardise retrieval, guardrails, and infra so product teams can ship safely and fast.
Key responsibilities
Stand up inference endpoints & routing across OSS/commercial models; manage prompt/versioning.
Build RAG pipelines (chunking, embeddings, vector DB) with caching, evals, and latency/cost SLOs.
Implement guardrails (PII redaction, jailbreak resistance, toxicity) and red-team tests; wire up hallucination/grounding metrics.
Create golden-path IaC (Terraform) modules; integrate with CI/CD, KMS/secrets, and tracing/cost observability.
Benchmark models/GPU utilisation; support fine-tuning/LoRA when justified by ROI.
Partner with Security/Compliance on PDPL/data-residency and audit readiness.
Candidate profile
5–8+ years in ML/Platform; 2–3+ years shipping LLM apps in production.
Strong Python + one typed language (TypeScript/Go/Java); containers/K8s; vector DBs (Pinecone/Weaviate/FAISS).
Experience with eval/guardrail frameworks and model observability.
Clear communication; comfortable owning end-to-end delivery.