SR Agentic AI Engineer

Position Summary Assent is building a governed agentic AI platform that takes use cases from business idea through to production — deployed as governed services accessible across the company.

You’ll work directly with business stakeholders across Sales, Customer Success, Enterprise Systems, and other functions to identify high-impact opportunities, design and build the agents that solve them, and ship them into a governed catalog where they become the new baseline for the rest of the organization.

This is not a classical ML role and not a traditional software engineering role.

You’ll spend your time designing agent loops, orchestrating tools and sub-agents, building RAG pipelines grounded in real enterprise data, and working with non-technical stakeholders to translate a business problem into a production-ready agentic solution — with observability, cost control, and governance built in from day one.

Key Requirements & Responsibilities Responsibilities include: Design and build production agent systems: tool calling, sub-agent orchestration, memory and context management, multi-step reasoning workflows — using frameworks such as LangChain, LangGraph, CrewAI, or MCP-based architectures.

Partner directly with business unit stakeholders to scope use cases: understand the problem, identify the right pattern, and co-design the solution rather than working purely from a written spec.

Build and tune RAG pipelines grounded in Assent’s enterprise data: Confluence, Jira, Snowflake, Salesforce — including retrieval strategy, embeddings, and evaluation for relevance and faithfulness.

Build evaluation harnesses and observability into every agent you ship: accuracy, hallucination rate, latency, and cost, with monitoring that holds up in production.

Take solutions from proof-of-concept to production deployment, working within established patterns for authentication, cost metering, and logging.

Contribute reusable patterns back into a shared library so that what you build for one team becomes accessible to the next — every solution should raise the floor for the whole organization.

Work closely with the platform and integration team to ensure what you build deploys cleanly into the governed catalog rather than living as a standalone one-off.

Your Knowledge, Skills and Abilities We strongly value your talent, energy and passion.

It will also be valuable to Assent if you have the following qualifications Required 5+ years of software engineering experience, with at least 2 years building production LLM-powered or agentic systems — not POCs or portfolio projects, shipped and operated in production.

Hands-on experience with at least one agentic framework: LangChain, LangGraph, CrewAI, AutoGen, or MCP-based orchestration.

Strong Python fundamentals; comfortable building APIs (FastAPI or similar) that expose agent capabilities to other systems.

Experience with RAG architecture end to end — chunking, embeddings, vector databases (Pinecone, Qdrant, Chroma, or similar), and retrieval quality tuning.

Cloud-native deployment experience on AWS (ECS, Fargate, Lambda, Bedrock) — you don’t need to own infrastructure, but you need to ship into it comfortably.

Demonstrated ability to work directly with non-technical stakeholders to scope and define a use case — this is as important as the technical skill set.

A clear point of view on what makes an agentic system production-ready: observability, cost control, failure handling, and a real evaluation methodology, not just a working demo.

Strong plusses Experience with LLM observability and evaluation tooling: Langfuse, LangSmith, Arize, or MLflow.

Experience with MCP (Model Context Protocol) or A2A agent-to-agent communication patterns.

Background contributing to or designing reusable architecture patterns adopted across teams — you think in patterns, not just point solutions.

Experience in a small or fast-moving team where you helped establish practices rather than inheriting them.