We're Hiring | AI Platform Engineer | Build What's Next in AI | Noida

We're Hiring | AI Platform Engineer | Build What's Next in AI | Noida

17 Nov
|
CLOUDSUFI
|
Noida

17 Nov

CLOUDSUFI

Noida

About the Role
We are building a unified AI platform for our Client. It lets every engineering team build, test, deploy and run AI agents and applications through standard paved paths, so they don't have to set up infrastructure, CI/CD, security or observability themselves. The platform is built on Google Cloud's managed agent ecosystem (Gemini Enterprise Agent Platform / Vertex AI).
Today it serves Google ADK agents running on Agent Runtime, with MCP tools hosted on Cloud Run. As an AI Platform Engineer, you will build the platform itself: the golden-path templates, SDKs, pipelines, governance, and telemetry that let a product team go from "I need an agent"; to "It's in production, secure and observable in hours instead of weeks. What the Platform Covers:
- Self-service onboarding: one reviewed config turns into a provisioned GitHub repo,CI/CD pipelines, webhooks, a baseline agent instance and access policies.
- Golden-path templates and an internal SDK: a standard agent repo layout (agentcode, prompts, evals, tests, Dockerfile) and a versioned Python SDK/CLI that holds the platform-owned logic.
- Agent CI/CD: pre-merge quality and security gates, smart diff detection (agent vs.prompt changes), immutable artifact promotion dev → test → stage → prod, and approvalgates.
- Prompt lifecycle: Prompts are versioned in Vertex AI Prompt Management, deployed separately from container builds, and fetched dynamically with caching.
- Self-service RAG infrastructure: standard, on-demand provisioning of document ingestion, embeddings, vector search and retrieval, so teams can ground their agents in enterprise data without building that pipeline themselves.
- Agent evaluation: golden datasets, LLM-as-a-judge and trajectory/tool-call evals, and post-deployment eval jobs on Cloud Run that report back to the pipeline.
- Security and governance: Agent Identity (per-agent SPIFFE identity), Agent Gateway(ingress/egress), Agent Registry, access policies with default-deny tool access, Model Armor, IAM/PAM and org-policy guardrails, all managed as Terraform.
- Observability: Conversation and token analytics in BigQuery (ADK BigQuery Analytics plugin), custom telemetry, Cloud Logging/Monitoring, and cost attribution per use case.
- Model abstraction: model routing and abstraction strategies so teams aren't locked into one model or version. Role Summary You will be the technical lead responsible for building a unified AI platform for one of our clients. You will own the platform architecture,



define the paved paths the client's engineering teamsuse to ship AI agents, and ensure the platform is secure, governed, observable and cost-efficient at enterprise scale. You will set engineering standards and work closely with the client's architects, cloud infrastructure, SRE and QA teams.

Key Responsibilities
- Own the platform architecture and roadmap: the reference architecture for agent development, deployment and runtime on GCP, across golden-path templates, SDK,CI/CD, evaluation, RAG infrastructure, governance and observability.
- Design the paved paths: a standard agent/prompt/tool repo model, the versioning strategy, and a developer workflow that keeps business logic separate from platform-owned code, enabling self-service onboarding that continuously improves speed.
- Drive onboarding automation: design a config-driven, single-approval flow (e.g., aGitOps/control-plane approach) that provisions repos, pipelines, webhooks, infrastructure and access policies across multiple IaC repos with no manual steps.
- Define the agent security and governance model: Agent Identity, Agent Gateway ingress/egress, Agent Registry, access policies (dry-run → enforce), Model Armor, org-policy exceptions, PAM/JIT access, and least-privilege IAM. Present these designs tothe client's security and cloud governance reviewers and get them approved.
- Set the agent quality strategy: decide which evaluations run in CI and which run afterdeployment, the eval-as-a-service architecture, quality thresholds and release gates,and regression protection for prompt and model changes.
- Lead the model abstraction and runtime strategy: model routing, versioning and fallback, context caching, prompt management, memory, and multi-agent patterns(A2A).
- Build the observability and FinOps foundation: a telemetry standard, BigQuery-based analytics, token and cost attribution per use case, SLOs and alerting, and closing observability gaps for onboarding teams.
- Lead architecture reviews (ARB), write design docs and decision records, and run POCs that compare build-vs-buy options.




- Mentor and grow L1/L2 engineers: code and design reviews, engineering standards forPython, Terraform and pipelines, and technical hiring.
- Work with client stakeholders (product teams, cloud infra, SRE, QA) to prioritise platform work, drive adoption, and represent the platform in technical forums. Must-Have

Skills:
Five years in software, platform, DevOps/SRE or AgentOps engineering, with 2+ years designing and building an internal developer platform or paved paths used by multiple engineering teams.
- Deep GCP expertise: IAM and org policies, resource hierarchy, VPC/PSC networking,Cloud Run, Vertex AI, BigQuery, Artifact Registry, Workload Identity Federation, Cloud Logging/Monitoring.
- Proven experience architecting and shipping GenAI/agent systems to production,including tool calling, MCP (Model Context Protocol) integrations, prompt management,evaluation, guardrails and multi-agent patterns.
- Expert-level Terraform (module design, multi-repo/multi-environment layouts, policy-as
- code, org-level IaC workflows).
- Has designed enterprise CI/CD: reusable templates, artifact promotion, approval gates,progressive delivery, pipeline security (Harness, GitHub Actions, GitLab or similar).
- Strong Python and API/SDK design skills. You can set code standards and review complex code.
- Robust foundation in cloud security: zero-trust, workload identity, secrets management, egress control, auditability.
- Track record of technical leadership: writing architecture documents, influencing without authority, mentoring engineers, and handling reviews with security and infra governance teams.
- Excellent written and verbal communication with senior stakeholders. Good-to-Have
- Hands-on with Google ADK and Gemini Enterprise Agent Platform (Agent Runtime, Agent Gateway, Agent Registry, Agent Identity, Memory Bank, Prompt Management).
- A2A protocol, Apigee, Model Armor, and AI-specific threat models (prompt injection, tool abuse, data exfiltration).
- Platform tooling such as Backstage, Crossplane, or GitOps control planes (ArgoCD/Flux).
- LLM evaluation frameworks (Vertex AI Gen AI Evaluation, ADK eval, DeepEval, Ragas) and LLM observability (OpenTelemetry GenAI conventions, Langfuse, Arize).
- Classic MLOps on GCP: Vertex AI Pipelines, Model Registry, Feature Store, BQML.
- Kubernetes/GKE experience.
- GCP Professional Cloud Architect / Professional Cloud DevOps Engineer certification.

📌 We're Hiring | AI Platform Engineer | Build What's Next in AI | Noida
🏢 CLOUDSUFI
📍 Noida

Reply to this offer

Impress this employer describing Your skills and abilities, fill out the form below and leave Your personal touch in the presentation letter.

Subscribe to this job alert:

Get the latest job offers by email for: we're hiring | ai platform engineer | build what's next in ai | noida / noida

Subscribe to this job alert:

Get the latest job offers by email for: we're hiring | ai platform engineer | build what's next in ai | noida / noida