5+ years of experience in building production software using modern programming languages.
Hands-on experience with AI coding tools like Claude Code or Cursor agent mode.
Ability to break down complex problems into precise specifications.
Strong understanding of testing strategies and debugging skills.
Proficient in reviewing code for correctness and performance, including code authored by others.
Comfortable evaluating non-deterministic outputs for quality.
Excellent communication skills and ability to collaborate with cross-functional teams.
Responsibilities
Turn product and technical intent into precise specs for agents to execute.
Break features and bugs into manageable tasks for agents.
Run multiple coding agents in parallel and keep them on track.
Build tests and checks to ensure agents self-correct before deployment.
Critically review agent-generated code for security and performance.
Maintain shared context for agents to improve overall system efficiency.
Establish coding standards and coach the engineering team on best practices.
Benefits
Opportunity to shape the future of payment infrastructure.
Work in a small, innovative team with a high degree of autonomy.
Engage in cutting-edge technology at the intersection of AI and software engineering.
Collaborate closely with product, design, and data science teams.
Contribute to the development of security and compliance standards in fintech.
Full Job Description
About the role
Want to be at the frontier of how the products of tomorrow get built? We're hiring Agentic Engineers who build by directing AI coding agents and own the outcome. Straddle is building the rails for the next era of payments, with the goal of moving money as fast as data moves. Identity verification, open banking, and compliance in a single API. You'll join Agentic Engineering, a new team at Straddle dedicated to building this way. We work alongside Product Engineering, Platform Engineering, and Data Science on the same roadmap, but a level up from traditional coding: turning ambiguous problems into precise specs, breaking them into agent-sized tasks, running several coding agents in parallel, and reviewing what they ship with a high bar. You'll write code by hand when that's the faster path, and delegate the rest. Because this is the infrastructure money runs on, security and correctness are never optional, and the bar you hold on agent output is what keeps it that way. The team is small and early, so you'll also shape how Straddle builds with agents: the standards, tooling, and shared context the rest of engineering will run on.
What you'll do
Spec before you build. Turn product and technical intent into precise specs agents can execute: clear interfaces, scope, and acceptance criteria. This is the highest-leverage part of the job.
Decompose for agents. Break features, migrations, and bugs into agent-sized tasks. Knowing what's too big for one pass, and how to slice it, is the craft.
Direct and orchestrate. Run multiple agents in parallel across features, integrations, and backlog. Keep them unblocked, correct course early, reset when a session drifts.
Own verification. Build the tests, builds, and checks that let agents self-correct. On payment and compliance paths, if you can't verify it, it doesn't ship.
Review at volume. Read agent diffs critically for correctness, security, performance, and fit. You own the merge, the deploy, and the consequences.
Improve the system. Maintain the shared context agents rely on (project memory, skills, hooks, integrations, CI) so everyone's agents get better. Treat specs and agent config as version-controlled engineering.
Keep it secure and compliant. Apply security best practices to agent-generated code: authn/authz, encryption, common-vulnerability defense, and fintech controls (PCI, SOC 2).
Set the standard. Establish conventions and coach the wider engineering team on specification, verification, and reviewing code they didn't write by hand.
Qualifications
What we're looking for
Strong engineering fundamentals. 5+ years building production software in a modern language, with the judgment to tell solid design from fragile design. That judgment is what makes directing and reviewing agents effective.
Hands-on agentic experience. You've shipped real work with tools like Claude Code or Cursor agent mode, well beyond autocomplete.
Specification and decomposition. You break down hard problems and write them down precisely.
Verification mindset. Fluent in testing strategy and making correctness machine-checkable, with sharp debugging and root-cause skills.
High-rigor review. You review fast and well, including code you didn't author, without dropping the bar.
At home with non-determinism. You evaluate output quality instead of assuming it.
Communication and teamwork. Strong writer, collaborates well with product, design, and data science.
Nice to have
Computer science foundations. A real understanding of data structures, algorithms, and complexity. No specific degree required.
Fintech or payments. Payment processing, banking software, ledgers, ACH, or software compliance.
Agent tooling or evals. You've built internal tooling, eval harnesses, prompts, skills, or CI that made a team better with AI.
Our stack. Less important than your judgment, as long as you're comfortable reviewing it as agent output: .NET (C#), TypeScript, and PostgreSQL, with solid architecture instincts (CQRS, Clean Architecture, DDD, SOLID).
Cloud. Azure, AWS, or GCP, architected for scale and reliability.
Leadership. You've led projects or helped a team adopt a new way of working.
