17 Nov
|
Deloitte Shared Services India
|
Mumbai
17 Nov
Deloitte Shared Services India
Mumbai
AGENT OPS ENGINEER
Location: India
Experience: 8-12 Years
Key Responsibilities
1. Agentic Application Deployment & LLMOps
- Design and implement deployment strategies specifically for Google ADK (Agent Development Kit) and GenAI agentic workflows.
- Establish LLMOps/FMOps pipelines to handle agent versioning, prompt registry management, and runtime configuration.
- Orchestrate agent runtimes, vector database integrations, and memory stores on scale-to-zero compute systems like Cloud Run or orchestration layers like GKE.
- Optimize latency, cold-start times, and compute costs associated with hosting agentic architectures.
2. GCP Cloud Infrastructure & Networking
- Configure and maintain high-performance, secure compute environments using GKE and Cloud Run optimized for AI workloads.
- Design secure VPC Networking topologies to support agent communication with internal APIs, external tools, databases, and Google Gemini models.
- Optimize BigQuery infrastructure for storing, indexing, and querying structured data or agent execution logs/telemetry.
3. Security, Guardrails, and IAM
- Implement strict GCP IAM policies to manage access controls for LLM APIs, agent tools, and data stores.
- Secure agent runtime environments using VPC Service Controls, Private Service Connect, and workload identity federation.
- Establish guardrails to monitor and secure egress calls made by autonomous agents.
4. SDLC Automation & GitHub Pipelines
- Build robust GitHub Actions pipelines designed for Agentic SDLCs, including automated testing of agent tools, prompt evaluations (evals), and deployment of agent manifests.
- Automate the packaging (Docker containerization)
and deployment of ADK-based applications to GCP Artifact Registry.
- Implement automated rollback and canary deployment strategies for live agentic applications.
Required Technical Skills and Qualifications
AI/Agentic Deployment Domain:
- Google ADK / Vertex AI Agent Builder: Direct experience deploying, scaling, and managing agentic applications built using Googles Agent Development Kit, Vertex AI Agents, or Dialogflow CX.
- LLMOps Concepts: Understanding of prompt tracking, agent tool integration, embeddings, and vector search deployment on GCP.
Core GCP Infrastructure:
- Containers & Serverless: Deep expertise in Cloud Run and GKE (with experience hosting AI middleware, API gateways, or agent runtimes).
- Data & Storage: Solid understanding of BigQuery and vector database integrations (e.g., Vertex AI Vector Search, pgvector, etc.).
- Networking: Practical knowledge of VPCs, Private Google Access, Load Balancing, and Secure API gateways.
- Security: Mastery of GCP IAM, Secret Manager, and service-to-service authentication.
DevOps & SDLC Tools:
- CI/CD: Advanced automation experience with GitHub Actions for complex, multi-stage deployment workflows.
- IaC: Hands-on experience writing Terraform configurations to provision resources for GenAI applications.
- Containerization: Mastery of Docker, container optimizations, and secure artifact storage.
Preferred Qualifications
- Experience setting up continuous evaluation (GenAI Evals) pipelines within CI/CD.
- Google Cloud Certified Professional Cloud DevOps Engineer or Skilled Machine Learning Engineer certification.
- Familiarity with Python-based agentic frameworks (e.g., LangChain, LangGraph,) and how they compile/deploy into the Google ADK ecosystem.
📌 GCP Agent OPS Engineer (Mumbai)
🏢 Deloitte Shared Services India
📍 Mumbai