Software Engineer - Agentic Code Review (Gitar)

About the role

Who is Sonar?

Sonar enables organizations to ship agentic code with confidence. Code creation is abundant, but confidence is scarce. Our core platform, SonarQube, verifies over 750 billion lines of code every day, delivering agentic verification and governance for all code. Trusted by more than 7 million developers and 75% of the Fortune 100, including teams at Nvidia, ServiceNow, Booking.com, Goldman Sachs, AstraZeneca, and Ford Motor Company, Sonar enables organizations to reduce outages, improve security, and lower the cost and risk of agentic software development.

Founded in Geneva and now headquartered in Austin, Texas, Sonar has a global team working across offices and hubs, including Silicon Valley, London, Bochum, Barcelona, Singapore, Tokyo, and the Washington, D.C. area. At Sonar, we’re enabling enterprises to build agentic software factories that can move quickly while maintaining the control, consistency and evidence required to ship with confidence.

Position description We are looking for a Software Engineer to join the team building Gitar's core platform. Gitar is building an AI-native code validation platform that automates the entire pull request lifecycle. We review every change with full awareness of the codebase, team conventions, and the intent behind the code; we generate fixes, validate them against CI, and only consider the work done when the build passes; and we automate the full pull request lifecycle from creation to merge.

You will work at the intersection of program analysis, developer tools, and AI, helping design and ship systems that reason about code at depth. You will do so by combining static and dynamic analysis with large language models to catch bugs that rules-based tools miss, generate validated fixes, and automate the developer's workflow end to end.

What you will do -

Design and build program analysis systems that reason about code at the level of intent, control flow, data flow, and behavior, across languages and at the scale of thousands of repos and large monorepos.

-

Combine static and dynamic analysis with large language models to detect functional bugs, logic errors, and behavioral issues that deterministic tools cannot catch.

-

Build the AI agents and orchestration layers that review pull requests, generate and validate fixes against CI, root-cause CI failures, and automate the full pull request lifecycle from creation to merge.

-

Develop indexing, retrieval, and embedding systems that give models the right context (codebase, conventions, dependencies, and intent) at the right cost and latency.

-

Own the production characteristics of the systems you build: correctness, accuracy, latency, cost, observability, and security in a zero-trust environment.

-

Partner with the SonarQube team to deepen the integration between deterministic static analysis and AI-native validation, building workflows where one informs the other.