SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Savvy is a Series C fintech company ($200M raised, $100M ARR in sight) reinventing wealth management through AI-native technology. The company partners with independent financial advisors, providing them a modern platform, investment management, and back-office operations so they can focus on client relationships and growth.
You will be Savvy's first dedicated agentic harness engineer, owning how developers across the engineering organization interact with coding agents and the systems supporting them. This is a hands-on platform engineering role focused on treating agent output quality and developer productivity as engineering problems, not prompt-engineering challenges.
Key responsibilities:
- Own the developer and agent experience: set roadmap for how engineers and coding agents navigate the codebase, use internal tools, run tests, review changes, and ship software.
- Build the cloud coding experience: develop the developer-facing layer around compute foundations, including task intake, environment setup, context, tools, permissions, observability, review, and delivery.
- Improve agent output at the source: change repository structure, interfaces, documentation, instructions, and feedback loops so agents produce correct code and sound architecture more consistently.
- Engineer context deliberately: design what agents receive upfront, discover progressively, persist, compact, hand off, or discard. Choose models and reasoning effort based on evidence.
- Own the internal skills system: raise quality of shared skills and instructions through contribution standards, evaluation, discoverability, maintenance, and pruning.
- Fix everyday developer friction: improve CI speed and reliability, local setup, testing, debugging, deploy workflows, and other bottlenecks.
- Measure what works: combine developer research and satisfaction data with DORA or SPACE measures and task-level agent evaluations. Establish baselines and test whether changes improve correctness, reviewability, cost, latency, and human intervention.
- Drive adoption: work directly with engineers, teach effective practices, and turn successful experiments into defaults.
- Partner on safe autonomy: work with infrastructure and security engineers on sandboxing, credentials, approvals, auditability, and controls for cloud agents.
You will join the infrastructure and harness team and partner closely with infrastructure, security, data, and product engineers.
REQUIREMENTS:
- Exceptional with coding agents: substantial hands-on experience using Claude Code, Codex, Cursor, or similar tools. Demonstrated ability to decompose tasks, manage context, choose models, inspect failures, and produce maintainable software. Output speed alone is not sufficient.
- Internal tools with real users: built developer tooling or a platform that engineers beyond your immediate team relied on. Learned from their behavior and support burden, not only from stated wants.
- CI or large-codebase experience: made meaningful improvements to CI, testing, repository workflows, or developer productivity in a substantial, long-lived codebase.
- Product judgment for internal systems: can interview developers, separate symptoms from root causes, turn evidence into a roadmap, and say no to attractive work that will not improve the system.
- Strong software engineering: built and operated substantial production software. Open on language and framework, but must be able to move from diagnosis to durable implementation independently.
- Evaluation over intuition: understand that agent behavior is nondeterministic and changing quickly. Compare prompts, context, tools, models, and workflows against representative tasks rather than declaring success from demos.
- Cross-team influence: can set standards and change defaults across teams without becoming a gatekeeper. Explain evidence, earn trust, and remain accountable for the experience created.
- Pragmatic range: comfortable moving between platform code, repository conventions, evaluation data, developer interviews, documentation, and hands-on enablement. This is not a prompt-writing, DevRel, or framework-only role.
NICE TO HAVE:
- Experience building cloud or remote development environments, coding-agent control planes, or secure execution sandboxes.
- Experience building or operating an agentic code review system.
- Experience designing agent tools, skills, context systems, memory, orchestration, or model-routing infrastructure.
- Experience using DORA, SPACE, or similar frameworks without reducing developer productivity to a single metric.
- Experience in financial services or another security- and compliance-sensitive environment.
