What You’ll Do
Design & Build with Google ADK: Design, develop, test, and deploy autonomous and semi-autonomous AI agents using Google ADK, Vertex AI Agent Builder, and agentic patterns to address key business challenges.
Vertex AI Implementation: Leverage Vertex AI capabilities including model training, fine-tuning, vector search, and orchestration to build production-grade AI solutions that integrate seamlessly with enterprise systems.
End-to-End Development: Own the end-to-end technical lifecycle of AI projects on Google Cloud Platform, from initial proof-of-concept and prototyping through to production deployment using Vertex AI Pipelines, monitoring with Vertex AI Model Monitoring, and continuous iteration.
Google Cloud Integration: Build integrations between AI agents and Google Cloud services including Gemini Enterprise, Cloud Functions, Cloud Run, Pub/Sub, and enterprise data sources to enable intelligent automation across the organization.
Cross-Functional Collaboration: Partner with business stakeholders to understand requirements and collaborate closely with QA Specialists and Cloud Engineers to ensure solutions are reliable, scalable, and seamlessly integrated within GCP infrastructure.
Responsible AI Implementation: Implement, test, and monitor safeguards for AI agents using Vertex AI's responsible AI tools to mitigate risks, including bias detection, model explainability, compliance monitoring, and operational limits. Ensure solutions are fair, explainable, and robust.
Agentic Architecture with ADK: Contribute to the architecture and selection of frameworks, tools, and orchestration logic using Google ADK and Vertex AI to build, manage, and scale multi-agent systems effectively.
Prototype Fast, in the Open: Ship a working prototype in days, not quarters. Put it in a real user’s hands early, watch them use it, and iterate on what you observe instead of what was requested.
Build MCP Servers and Tool Integrations: Design, build, and maintain Model Context Protocol (MCP) servers, connectors, and tool layers that give Claude, Gemini, and other models safe, governed, audited access to Coupa business systems and data.
Stay Multi-Model and Pragmatic: Choose the right model, framework, and pattern for each job across Claude, Gemini, and open models. Make cost, latency, and accuracy tradeoffs deliberately instead of defaulting to one vendor.
Know When It Isn’t an Agent: Recognize when workflow automation, a scheduled job, a cleaner integration, or simply deleting a step is the faster, cheaper, more durable fix. Reach for the smallest tool that solves the problem.
Instrument and Prove Impact: Measure every deployment. Track usage, cycle-time reduction, hours returned, and error and escalation rates, then report results to stakeholders and leadership in business terms, not model metrics.
Carry Trust and Governance into the Field: Apply security, privacy, data-handling, and responsible AI guardrails at the point of deployment. Engage Security, Legal, and Data Governance early so solutions clear review the first time.
Enable, Document, and Hand Off: Leave behind runbooks, monitoring, and trained business owners so what you build survives without you. Convert one-off wins into reusable patterns, templates, and internal skills the whole AI Delivery team can redeploy.
Iterate & Improve: Monitor agent performance in production using Vertex AI monitoring tools, gather feedback, and iteratively improve models, prompts, grounding strategies, and agent logic to drive successful adoption and business impact.
What You Will Bring to Coupa
Proven experience (3-5+ years) in software engineering, data science, or AI/ML solution delivery with demonstrated Google Cloud Platform expertise.
Hands-on experience with Google Vertex AI, including model deployment, endpoint management, and AI agent development.
Strong knowledge of AI/ML fundamentals, agentic AI patterns (e.g., ReAct, planning, tool use, function calling), and experience with Google ADK or Vertex AI Agent Builder.
Proficiency in Python and Google Cloud AI/ML libraries including Google ADK, Vertex AI SDK, and common data science libraries (e.g., pandas, scikit-learn, TensorFlow/JAX).
Experience with Google Cloud Platform services (Vertex AI, BigQuery, Cloud Functions, Cloud Run, Pub/Sub) and MLOps principles including Vertex AI Pipelines and CI/CD.
Demonstrated success in building and deploying AI/ML models and solutions into production on cloud platforms.
Demonstrated experience in a forward deployed role where you owned the business outcome, not just the ticket.
Comfort operating in ambiguity. You can walk into an undefined business problem, scope it with the people who own it, and ship something useful within weeks.
Experience integrating AI or automation into enterprise SaaS platforms (ERP, CRM, HRIS, procurement, ITSM) through REST APIs, webhooks, event streams, or middleware.
Strong written communication. You can explain a cost and accuracy tradeoff to a CFO and a stack trace to an engineer on the same afternoon.
Working knowledge of at least one non-Google model provider and its tooling (e.g., Anthropic Claude, OpenAI), including function calling and tool use across providers.
Proven ability to communicate complex technical concepts to both technical and non-technical stakeholders.
Bachelor's degree in Computer Science, Data Science, or a related field, or equivalent practical experience.