Senior Software Engineer

What you'll do

• Design and implement a multi-agent hierarchy using GCP Agent Development Kit (ADK) with various agent classes
• Write and tune system prompts, tool schemas, and output schemas for reliable agent behavior
• Build Python tool layer connecting agents to BigQuery, Vertex AI Search, Vertex AI prediction endpoints, Meta Marketing API, Pub/Sub, and Secret Manager
• Implement multi-mode orchestration: chatbot, advisor, and comparator modes
• Design session memory and conversation context management backed by Cloud Firestore
• Own prompt engineering lifecycle including writing, version control, A/B testing, and iteration
• Maintain golden evaluation set and LLM-as-judge evaluation pipeline to ensure quality thresholds
• Implement ReAct loops, self-correction, and reflection loops to improve answer quality
• Establish guardrails for input classification, output validation, and PII scrubbing
• Deploy conversational agents to Vertex AI Agent Engine and infrastructure agents as Cloud Run Jobs
• Build and maintain CI/CD pipelines using Cloud Build with evaluation gates and deployment promotion
• Configure IAM service accounts with minimum required permissions
• Instrument agents with structured Cloud Logging and set up Cloud Monitoring dashboards and alerts
• Own BigQuery tool layer: write and optimize SQL queries, manage gold dataset schemas, ensure query cost efficiency
• Maintain and grow Vertex AI Search knowledge base
• Collaborate with ML team to integrate Vertex AI prediction endpoints into agent tools
• Design and populate Config Store tables in BigQuery used by ComparatorAgent
• Define agent engineering standards including code structure, tool schema conventions, prompt versioning, eval thresholds, and deployment gates
• Conduct code and prompt reviews and provide technical mentorship
• Translate marketing requirements into well-scoped agent capabilities and write technical specs
• Identify capability gaps and drive resolution across engineering teams

Requirements

• 5+ years of software engineering experience, with at least 2 years in production LLM or AI systems
• Experience building multi-agent systems with agent hierarchies, tool use, ReAct loops, self-correction patterns in production
• Strong prompt engineering skills including system prompt design, few-shot example construction, output schema enforcement, and structured output parsing
• Hands-on experience with at least one agentic framework such as GCP ADK, LangChain, LangGraph, CrewAI, AutoGen or equivalent
• Understanding of LLM evaluation methods including golden datasets, LLM-as-judge, RAGAS or similar metrics, and A/B testing
• Production experience with Vertex AI including model endpoints, Vertex AI Search, and ideally Vertex AI Agent Engine or Reasoning Engine
• Strong BigQuery skills including complex SQL, partitioning and clustering strategy, query cost optimisation, and schema design for analytics workloads
• Experience with Cloud Run, Cloud Functions, Cloud Scheduler, Pub/Sub, and Cloud Build for production workloads
• Familiarity with GCP IAM, Secret Manager, Cloud Logging, and Cloud Monitoring
• Python proficiency including async programming, Pydantic models, FastAPI or similar frameworks, and clean modular code structure
• Strong understanding of REST APIs, JSON schema design, and API integration patterns
• Experience with Docker, container registries, and CI/CD pipelines
• Git-based development workflow including branching strategy, PR reviews, and automated testing in CI

Tech stack