Job Description & Details
This is a heavy-duty GenAI role focused squarely on building autonomous, agentic workflows using Google Cloud infrastructure. You'll be bridging the gap between raw enterprise data in BigQuery and cutting-edge LLMs to ship real production systems that actually drive business automation.
What You'll Actually Be Doing
Your day-to-day will revolve around architecting and deploying multi-agent systems using Vertex AI Agent Builder and the Agent Developer Kit. You won't just be playing with prompts; you'll be writing robust Python pipelines, hooking up vector engines in AlloyDB or BigQuery for real-time grounding, and ensuring your models pull live context without hallucinating. Expect to spend a lot of time engineering data flows and optimizing how these intelligent agents securely interact with massive corporate databases.
The Core Tech Stack
You need deep, hands-on experience with Vertex AI (Model Garden, Pipelines, Agent Builder) and an advanced grasp of Python and SQL. Google Cloud is the non-negotiable playground here—specifically knowing your way around BigQuery, Dataflow for streaming pipelines, and Cloud Storage. If you haven't worked with vector search engines for RAG or agentic frameworks like Model Context Protocol, you'll find yourself struggling to keep up with the architecture demands.
Interview Expectations
Expect the hiring team to grill you on how you handle state management and error recovery in multi-agent workflows when an upstream LLM call fails or returns garbage. They'll also likely push you on optimization strategies for real-time inference over massive datasets in BigQuery, looking to see if you understand the actual latency and cost trade-offs of streaming embeddings.
Application Advice
Don't just list Python and GCP on your resume; explicitly highlight projects where you deployed Vertex AI endpoints, built RAG pipelines, or orchestrated LLMs with custom data tools. Make sure keywords like Vertex AI Agent Builder, BigQuery, Dataflow, and vector embeddings stand out so you clear the initial ATS filters without getting lost in the noise.