Cloud & AI Platform Architect

We are looking for one architect to own both layers. Your foundation is deep platform and DevOps architecture: reliability, scalability, security, and cost efficiency of the cloud platform every R&D team builds on. Your growth edge is the AI layer: designing how agents are orchestrated, how they access tools and data safely, how we evaluate whether they work, and how we govern their cost and behavior in production.

We know agentic AI is a young discipline. We are not looking for a decade of AI experience that does not exist. We are looking for a proven platform architect who has spent the last one to two years genuinely building LLM-based and agentic systems in production, and who wants to own where this field goes inside a company that takes it seriously.

What Youll Do:
Platform & DevOps
Cloud Architecture: Own the architecture of our cloud platform, including GKE fleet design, VPC networking, IAM/Workload Identity, and multi-environment strategies across GCP (and AWS where relevant).
IaC & GitOps: Lead Infrastructure-as-Code (Pulumi, Terraform) standards, ArgoCD deployment patterns, and secure CI/CD paved paths (GitHub Actions, GitLab CI) adopted by all R&D teams.
Reliability & Observability: Own reliability architecture, including Disaster Recovery (DR) strategy and drills, P1 incident reduction, MTTR improvement, and observability platform architecture (Datadog, Prometheus, Grafana, OpenTelemetry) including usage and cost optimization.
FinOps: Drive FinOps as an architectural discipline-rightsizing, idle-resource elimination, unallocated-spend attribution, and cost-aware design reviews using tools like Kubecost or GCP Cost Management.
Agentic AI

Agent Architecture: Design our agentic AI ecosystem, focusing on agent orchestration frameworks (e.g., LangChain, LangGraph, CrewAI, AutoGen), tool/MCP (Model Context Protocol) interfaces, vector databases/memory strategies (e.g., Pinecone, Qdrant, pgvector), and production AI deployment patterns.
Evaluation & Guardrails: Build the evaluation and guardrail layer for AI in production using frameworks like Ragas, TruLens, or LangSmith. Define permission models, human-in-the-loop checkpoints, audit logging, and policy-as-code for AI usage.
AI FinOps & Monitoring: Own AI cost observability, token spend monitoring, model routing strategies (optimizing for cost/quality/latency trade-offs), and real-time alerting on runaway model usage.
Reference Architectures & Standards: Define reference architectures for teams building AI features, RAG patterns, prompt versioning/management, and model API standards (OpenAI API, Anthropic, open-source models via vLLM/Ollama), driving high-leverage AI automation across R&D.