Software Engineer - Agentic Development at Fair Isaac in Work from Home

Job description

FICO (NYSE: FICO) is a leading global analytics software company, helping businesses in 100+ countries make better decisions. Join our world-class team today and fulfill your career potential! The Opportunity FICO builds the analytics and decisioning technology behind some of the highest-stakes decisions in the world economy including the FICO Score, the standard measure of consumer credit risk in the US, and FICO Platform, an API-first applied intelligence platform delivering real-time decisions across fraud detection, credit risk, regulatory compliance, marketing, and customer engagement. This is mission-critical, low-latency, heavily regulated software where auditability and controlled change are non-negotiable. This role sits at the point where that technology is being built and operated in an agentic-first way. As coding agents take on an increasing share of the software lifecycle, the engineer's role shifts from hand-writing code to directing, specifying, and verifying the output of a fleet of specialised agents, working through a structured harness of guides, feedback loops, guardrails, and shared context that transforms raw model capability into production grade engineering. The successful candidate will bring the engineering judgement that agents cannot replicate: the ability to define what good looks like, encode it with sufficient precision for agents to achieve it, and recognise when they have not, applying that judgement to architecture, correctness, trade-off analysis, and the design of the work itself. What You’ll Contribute Collaborate with Product teams to translate early stage ideation into structured Proof of Concepts, establishing clear technical feasibility and scope. Produce detailed specifications and acceptance criteria that enable coding agents to generate correct, maintainable, and well-architected output, taking full ownership of the resulting code through to deployment. Apply rigorous quality standards to all agent generated output, ensuring it meets the same level of scrutiny as hand written code within a regulated decisioning context. Operate effectively within the agentic harness, utilising established guides, including AGENTS.md files, architectural rules, and reference documentation alongside automated sensors such as linters, fitness tests, verification loops, and LLM-as-judge reviewers. Identify the root cause of recurring agent errors and work with the harness team to implement systemic controls, ensuring the same class of failure cannot recur across the codebase. Maintain the accuracy and currency of repository documentation, specifications, and contextual knowledge, treating these as a shared system of record for both human engineers and agents. Apply focused technical judgement to architecture decisions, edge case analysis, latency and integration risk, auditability requirements, and any determination that falls outside the appropriate scope of agent autonomy. What We’re Seeking Demonstrable experience in software engineering within large, complex codebases, with a strong foundation in software architecture, testing practices, and long-term maintainability. Practical experience working with AI coding agents such as Claude Code, GitHub Copilot, Codex, or Cursor with a clear understanding of their capabilities, limitations, and appropriate use within a production engineering workflow. The technical judgement to critically evaluate agent generated output, identify subtle errors or architectural weaknesses, and make informed decisions about when to intervene in the agent workflow. The ability to communicate technical requirements with precision, producing specifications and definitions of done that are unambiguous, testable, and actionable by both agents and engineering colleagues. A disciplined approach to verification, instinctively questioning correctness, building appropriate checks and test coverage, and applying particular rigour in regulated and mission-critical delivery contexts. A re