Location: Mid-Town NYC, USA
Work Model: In-person 3 days/week

Experience Level: 5+ Years

You will join a focused pod building two tightly-coupled capabilities for Humana's internal users:

1. Internal Agentic Search Engine

• Build agentic pipelines that plan, retrieve, and reason across numerous internal and enterprise data sources (structured and unstructured).

• Design robust retrieval, grounding, and orchestration layers so agents return accurate, cited, and trustworthy answers.

• Implement tool-use, multi-step reasoning, and evaluation loops that keep quality high as sources and scope grow.

2. Perplexity-Style Front End

• Deliver a fast, intuitive, conversational search experience for Humana employees — answer-first, with sources, follow-ups, and streaming responses.

• Partner across the stack to connect the agentic backend to a responsive, production-grade UI.

• Instrument usage, feedback, and quality signals to continuously improve relevance and user trust.

• Build hands-on, end to end. Design, code, test, and ship agentic AI features into production — you write code every day, not just architecture diagrams.

• Engineer agentic systems. Develop planning, retrieval, tool-use, and orchestration components for the internal agentic search engine.

• Integrate many data sources. Connect agents reliably and securely to numerous internal and enterprise data sources.

• Deliver on Google Cloud. Build, deploy, and operate solutions natively on GCP (Vertex AI, GKE, BigQuery, Cloud Run, and related services).

• Ship the user experience. Contribute to the Perplexity-style front end so end users get fast, grounded, well-cited answers.

• Own quality and evaluation. Establish evals, guardrails, observability, and feedback loops to keep answers accurate and safe.

• Collaborate on-site. Work in person at the NYC Mid-Town office Tuesday through Thursday, pairing closely with the pod and Humana stakeholders.

• Move fast. Operate with urgency in a fast-moving engagement — iterate quickly, unblock yourself, and drive outcomes.

• Strong, current software engineering fundamentals — clean, tested, production-quality code (Python strongly preferred).

• Demonstrated, hands-on experience building agentic AI systems (agents, tool-use, planning, multi-step reasoning, RAG/retrieval).

• Deep expertise with the Google Cloud technology stack — e.g., Vertex AI, GKE, BigQuery, Cloud Run, Cloud Storage, IAM.

• Experience integrating LLM applications with numerous, heterogeneous data sources (APIs, databases, document stores, search).

• Experience taking ML/GenAI systems to production — deployment, scaling, monitoring, and reliability.

• Ability to work on-site in NYC Mid-Town 3 days per week (Tuesday–Thursday).

• Breadth of knowledge across frontier models (e.g., Gemini, and other leading LLM families) and when to use which.

• Hands-on experience with open-source agentic and LLM frameworks (e.g., LangChain, LangGraph, LlamaIndex, or similar).

• Front-end / full-stack exposure to help deliver a Perplexity-style user experience.

• Experience with evaluation frameworks, guardrails, and responsible-AI practices.

• Prior experience in regulated or enterprise environments (healthcare a plus)

The right person for this team is a builder at heart — someone who is energized by shipping working software, thrives in ambiguity, and raises the quality bar for everyone around them.

• Extremely hands-on: happiest in the codebase, shipping real features.

• Bridges agentic AI and solid software engineering — not just prototypes, but production systems.

• Pragmatic and fast — comfortable with a high-tempo, fast-moving engagement.

• Strong communicator who collaborates well in person with a tight-knit pod and client stakeholders.

• Curious about the frontier — keeps up with new models, frameworks, and techniques.