Agentic DevEx Engineer

Ekho•New York, NY

About The Position

We're hiring an Agentic DevEx Engineer — a systems-minded engineer who owns the infrastructure, standards, and tooling that make AI agents effective collaborators in our codebase. This is a rare role at the intersection of Platform Engineering, Developer Experience, and AI. You'll be responsible for transforming how humans and agents work together at Ekho. Today, AI coding agents are powerful but inconsistent. They hallucinate file paths, miss context, violate conventions, and require constant hand-holding. Your job is to eliminate that friction. Singular Mandate: Maximize the effectiveness of agents inside Ekho’s codebase so that both humans and AI can execute predictably, safely, and at scale.

Requirements

5+ years as a software engineer, with significant experience in infrastructure, platform, or developer experience roles.
Strong full-stack fundamentals : you can work across frontend (React/TypeScript) and backend (Node.js, Firebase/GCP) comfortably.
Track record of improving developer velocity through tooling, automation, or infrastructure — not just shipping features.
Excellent systems thinking : you can design for maintainability, enforce boundaries, and see second-order effects.
Strong written communication : you can turn messy codebases into clear documentation and tribal knowledge into explicit standards.
Obsessively curious about AI tooling. You've used Claude Code, Cursor, Copilot, or similar extensively.
You have opinions about what makes agents succeed or fail.
You're excited to push the boundary of what's possible.

Nice To Haves

Experience building internal developer platforms or DevEx tooling from scratch.
Contributions to open-source AI coding tools or significant CLAUDE.md / AGENTS.md work.
Background in CI/CD systems , automated testing infrastructure, or static analysis tooling.
Familiarity with prompt engineering and how LLMs consume context.
Prior experience at a high-growth startup where you wore multiple hats and moved fast.

Responsibilities

Codebase Architecture & Structure Define and enforce architectural boundaries : establish clear domain separation, eliminate duplicated modules, and create discoverable code organization.
Build reproducible development environments : standardized commands ( dev , test , lint , build ) that work identically for humans and agents across all surfaces.
Maintain the single source of truth for how the codebase is organized, documented in ARCHITECTURE.md files per domain.
Agent Infrastructure & Tooling Own AGENTS.MD / CLAUDE.MD : transform our agent instructions from verbose, ad-hoc documents into concise, high-signal guides that improve agent accuracy.
Design and implement agent hooks : deterministic workflows for how agents plan, read context, execute changes, and validate output.
Build a library of reusable agent skills : well-defined, domain-specific capabilities (e.g., "analyze payment invoice", "update Firebase function safely") with clear contracts.
Ensure agents can navigate the codebase without tribal knowledge — predictable paths, explicit boundaries, and machine-readable context.
Standards & Enforcement Codify engineering standards : capture Ekho's conventions (formatting, naming, patterns, error handling) explicitly and enforce them through linters.
Improve the automated review layer : CI checks, static analysis, and AI-powered reviewers that catch misalignment before human review.
Own validation infrastructure : improve test reliability, reduce flakiness, establish schema guarantees, and ensure CI provides fast, accurate feedback.
Create the feedback loop: track recurring issues, refine standards, and continuously improve the system.
Developer Experience & Adoption Define work modeling standards : templates for tickets and specs that capture context, constraints, and acceptance criteria explicitly — so agents can execute without ambiguity.
Build canonical workflows : documented patterns for common tasks (bugfix, feature, refactor) that engineers follow with agent assistance.
Own AI observability : instrument engineering workflows to track PR cycle time, AI-assisted PR percentage, and other signals of agent effectiveness.
Drive adoption : train the team on agent collaboration patterns, create demos, and ensure the right way is the easy way.