Internal Tooling Engineer

Relay; France; Remote; EUR 90,000 - 130,000; Full time

Benefits offered by this job

Competitive salary

Equity package

Health benefits

Unlimited PTO

Remote-first culture

Job summary

Relay is seeking an Internal Tooling Engineer to boost productivity across the engineering org. You will ship backend tooling daily, build and instrument services, and act as a forward-deployed specialist improving developer workflows with AI agents.

You will own the shared toolset for coding agents, ensuring scalable, safe workflows, and reduce repetitive toil across teams. This is a hands-on IC role with no direct reports.

Qualifications

• Hands-on backend engineering with production services.
• 5+ years in software engineering, developer tooling, or platform engineering.
• Strong TypeScript and experience building internal tools and API integrations.
• Production experience with LLM agents: tool use, retrieval, orchestration, and evals for nondeterministic behavior.

Responsibilities

• Improve agent-centric development workflows.
• Own shared library of AI agent skills and context for engineering teams.
• Watch and fix AI-assisted workflows, tooling, or guidance.
• Track cycle time, token cost, and rework metrics.

Skills

Backend engineering

TypeScript

Internal tools

LLM agents

CI/CD

Datadog observability

Code review

Technical writing

Tools

Datadog

Job description

About Us

Relay is a cross-chain payments protocol that provides instant, low-cost cross-chain bridging, swaps, and transactions. We're the fastest and cheapest way to bridge and transact across chains, serving over 5 million users who have completed 59+ million transactions with $6+ billion in volume across 85+ chains.

About The Role

We're hiring an Internal Tooling Engineer with one goal: make every Relay engineer a lot more productive.

You'll work "inside the box". Your customers are our own engineers, and the things you build never ship to end users but decide how fast and how safely everything else ships. Think of the role as a forward-deployed engineer for our engineering org: you find where people wait, get stuck, or repeat themselves, and you remove it.

In the short term, this is an individual contributor, hands-on backend engineering role. You (or your agents) will write code every day, deploy it, and run it in production: building services and integrations, instrumenting them, setting up monitoring and alerts, and being on call for what you ship. There is no team to manage. If you want to lead people or mostly advise, this isn't the role.

What you'll do

Improve our agentic development workflows

• Work with our internal AI agent, Scout, and the teams that use it, to turn repetitive engineering work into agent skills: investigation, triage, ticket routing, status updates, and first-draft fixes
• Own the shared library of skills, tools, and context that engineers' coding agents rely on, so what works for one team works for all, whether they use Claude, Codex or any other flavor of a coding agent.
• Watch how engineers actually work with AI tools, find where they get stuck, and fix the workflow, the tooling, or the guidance
• Track cycle time, token cost, and rework, and report what each change moved

Make agents the first reviewer of all code

• Put in place mechanisms so agents review and potentially approve (or request changes!) every PR, and engineers are never blocked waiting for another human's review.
• Close the loop on every review: capture each human override, missed bug (false negative), and wrong or dismissed comment (false positive), and feed them back into evals, prompts, and rules, so the agent approves more and gets it wrong less over time
• Build the surrounding safety net: tests, deploy guards, and rollback, so faster review does not mean riskier releases
• Track false-positive and false-negative rates over time, and widen what the agent may approve on its own only as those rates fall

Make debugging issues trivial

• Work with tools like Datadog so the logs, traces, metrics, and recent deploys behind any issue are one query, or one agent question, away
• Improve instrumentation and alert ownership so every alarm points to a team and a likely cause
• Cut observability noise and cost without losing the signals engineers need
• Turn recurring incident patterns into runbooks and automation that Scout can run

About You

• You're a hands-on engineer who ships tools other engineers use every day, and you treat them as your customers
• You're obsessed with cycle time: you notice every minute someone waits, and you find out why
• You use AI coding agents heavily yourself and know what makes them reliable or flaky
• You'd rather add a test, a guardrail, or a lint rule than rely on a prompt
• You know the difference between an AI review comment engineers act on and one they learn to ignore
• You're an IC who's comfortable being the first dedicated hire in a new function, and who works well with people who are busy and skeptical
• You build feedback loops: you measure before and after, learn from every miss, iterate, and report honestly when something didn't help

Skills And Abilities

Must-have

• Hands-on backend engineering experience: you build, deploy, and maintain production services yourself, including instrumentation, logging, monitoring, and alerting
• 5+ years in software engineering, developer tooling, or platform engineering
• Strong TypeScript and experience building internal tools and API integrations
• Production experience with LLM agents: tool use, retrieval, orchestration, and evals for nondeterministic behavior
• Experience bringing AI-assisted code review or testing into a team's daily workflow, measuring the result, and improving it through feedback loops that reduce false positives and false negatives
• Working knowledge of CI/CD and code review systems, and what makes them slow
• Experience with observability tools such as Datadog: logs, traces, metrics, monitors, and cost control
• Strong written communication; you can turn an engineer's pain point into a precise spec

Nice to have

• Experience with Rust (we use a little of it)
• Experience with MCP or comparable tool and context protocols
• Experience designing risk-based merge policies, deploy guards, or automated approval flows
• Experience with incident tooling, on-call workflows, or runbook automation
• Experience with Slack apps or internal platforms
• Experience in crypto, payments, or another high-stakes domain

What success looks like

• First 30 days: you've shadowed engineers across teams, mapped where time is lost from commit to production, and agreed a baseline for cycle time and review wait
• First 60 days: agent review covers a first set of repositories, with clear rules for what merges without a human and a feedback loop capturing every override and miss
• First 90 days: median wait for a first review is measured in minutes, the review agent's false-positive and false-negative rates are tracked and falling, and a Datadog-backed debugging flow is in daily use
• Ongoing: engineers ship visibly faster, and the review agent keeps improving: it approves more on its own, with fewer false positives and false negatives each month

What we offer

• Competitive base salary
• Equity package
• Comprehensive health benefits
• Unlimited PTO policy with encouraged minimum
• Remote-first culture with emphasis on collaboration, communication, and flexibility