Senior Backend Engineer, AI Engineering: Chat

As a Senior Backend Engineer on the Duo Chat team, focused on Chat Engine, you'll build the core AI capabilities behind GitLab Duo Chat, the natural-language and agentic interface to the GitLab DevSecOps platform. Much of your work will involve building flow components and agentic flows in the Flow Registry, the Python framework built on LangGraph within our Duo Workflow Service that powers agentic chat, and integrating them with the GitLab Rails monolith.

You'll own complex backend features from start to finish across both services. You'll integrate large language models and orchestrate multi-agent flows so customers can work faster and more securely across the software development lifecycle. This work sits where GitLab's core platform meets its AI strategy, and reliability, performance, and answer quality directly shape the customer experience.

• Design and build flow components and agentic flows in the Flow Registry using Python and LangGraph within the Duo Workflow Service. These reusable building blocks power agentic Duo Chat and, increasingly, other AI features across GitLab.
• Develop, ship, and maintain backend features for GitLab Duo Chat across the Python Duo Workflow Service and the GitLab Rails monolith in a secure, well-tested, and performant way.
• Integrate new generative AI models, providers, tools, and multi-agent orchestration patterns into Duo Chat to expand its capabilities and improve answer quality.
• Design, implement, and review GraphQL and Representational State Transfer (REST) application programming interfaces (APIs) and related monolith logic, including chat entry points, permissions, and foundational-flow registration. Keep contracts with frontend clients and host systems reliable and clear.
• Improve debugging, observability, and test coverage using pytest, RSpec, and related frameworks; track and improve latency, error rates, and test coverage so AI-powered chat workflows stay reliable at scale.
• Collaborate with Product, User Experience (UX), frontend, and AI specialists to refine requirements and deliver high-quality improvements through iteration.
• Document standards, patterns, and learnings with other engineers, raising the bar for safe AI integration and evidence-driven engineering.
• Participate in Tier 2 on-call rotations to troubleshoot production issues, contribute to root cause analysis, and strengthen resiliency.

• Significant experience building and maintaining production Python backends, including APIs, data models, and asynchronous or long-running workloads.
• Practical experience designing and shipping AI-powered, agentic backend features, including large language model integration, tool or function calling, and multi-agent orchestration. You use sound judgment about large language model limitations and safe use in production.
• Working proficiency in Ruby on Rails, or a strong willingness to learn it. Duo Chat integrates deeply with the GitLab monolith for chat entry points, GraphQL, permissions, and flow registration, and most of GitLab's codebase is written in Ruby.
• Proficiency designing or extending REST or GraphQL APIs with attention to scalability, maintainability, and backward compatibility.
• Strong Structured Query Language (SQL) skills and familiarity with relational databases such as PostgreSQL, including efficient queries and data modeling.
• Ability to find, diagnose, and prevent performance and reliability problems at scale.
• Experience solving technical problems of high scope and complexity and advocating for quality, security, and performance improvements across your team.
• Openness to learning and collaborating in an async-first, distributed team, applying transferable skills from related technologies or domains. Hands-on experience with agent frameworks such as LangGraph or LangChain is a strong plus.

We're part of GitLab's AI Engineering organization and own the AI-powered chat experience embedded across the GitLab platform. Chat Engine is the backend-focused team within the Duo Chat group.

We're backend, frontend, and AI specialists working asynchronously across time zones, using issues, merge requests, and documentation as our main collaboration tools. Our focus is to expand generative and agentic AI capabilities, improve the performance and reliability of chat workflows, and strengthen the debugging and testing foundations that let us run AI features safely at scale.

For more on how we work, see the team handbook page.