Staff Fullstack Engineer, Agentic ApplicationsActive$192K
The opportunity
Databricks is transforming how it builds and operates People Technology — moving from traditional SaaS configuration toward an AI-native, agentic stack. You'll be the technical anchor of the People Tech pod, driving the architectural shift from workflow automation to autonomous,…
What you'll do
Architect and build agentic systems that automate and augment People Tech: workflows — onboarding, offboarding, comp analysis, policy Q&A, HR service delivery — using LLM orchestration frameworks (LangGraph, AutoGen, or equivalent).
Define the agentic platform strategy for the pod: agent design patterns, tool-calling conventions, retrieval-augmented pipelines, evaluation frameworks, and human-in-the-loop guardrails.
Integrate People Tech systems (Workday, Greenhouse, ADP etc.) as: agent-accessible tools and data sources via Databricks Unity Catalog and MCP-style interfaces.
Set the technical bar for the pod: reviewing designs, establishing engineering standards, and leading architectural reviews across the People Tech roadmap.
Influence peers and stakeholders: translate agentic capability into business outcomes for People, Legal, and Finance partners, and mentor engineers in the pod on AI-first thinking.
+ years of software engineering experience, with at least 2 years building: production LLM or agentic applications (agents, RAG pipelines, tool-use, multi-agent orchestration).
What they're looking for
- Deep fluency in Python and experience with agentic frameworks: LangChain/LangGraph, CrewAI, AutoGen, Semantic Kernel, or similar.
- Strong command of enterprise integration patterns: REST/GraphQL APIs, event-driven architecture, and connecting SaaS HR/HCM platforms programmatically.
- Experience with data platforms: Databricks, Spark, or equivalent — and building AI applications on top of lakehouse or warehouse architectures.
- Track record as a technical lead: driving architectural decisions, writing RFCs, and raising the quality bar across a team without relying on management authority.
