AI Software Engineer

Description

Job Summary: Implement features, orchestrate specialized subagents, and build Skills and MCP servers, maintaining ownership of the plan and final diff review. Key Highlights: 1. Production experience with coding agents 2. Proficiency in multiple programming languages 3. Building and maintaining Skills and MCP servers * Minimum 2 years of experience developing and maintaining production software. * Completed university degree in Systems Engineering, Computer Science, Statistics, Economics, Business Administration, Computer Sciences, or related fields. Essential Technical Skills * Proficiency in TypeScript/JavaScript, Python, Go, or similar — Advanced * Real-world experience using a coding agent (e.g., Claude Code, Codex, or equivalent) in production work—not just experiments. — Intermediate * Knowledge of Git, testing, API design, architecture, and code review. — Intermediate Desirable Technical Skills * Claude Code: project memory (CLAUDE.md), plan mode, slash commands, hooks, plugins, and context management. * Skills (SKILL.md): reusable instruction packages defining how to perform a task; automatic activation based on context. * Subagents / Agent Teams: specialized agents with their own context, prompts, and permissions, coordinated by a primary agent. * MCP (Model Context Protocol): an open standard enabling agents to access APIs, databases, and external services. * OpenAI Codex: terminal-based coding agent; criteria for combining it with Claude Code depending on the task. * AI-powered testing: test generation and execution, verifying that the agent actually runs and fixes real failures. * Quality and security: linters, type-checking, secret scanning, least privilege, and prompt injection defense. * Clear understanding of the differences among Skills, subagents, and MCP, and sound judgment in selecting the appropriate mechanism. * Ability to critically review AI output and maintain a security mindset (secrets, permissions, dependency validation). * Experience building custom MCP servers or team-shared Skills/plugins. * Familiarity with more than one agent ecosystem (e.g., Claude Code and Codex). * Knowledge of model evaluation (evals) and context/token cost management. Responsibilities * Implement full features by delegating bounded tasks to coding agents, maintaining ownership of the plan and final diff review. * Translate product requirements into agent-executable plans: task decomposition, acceptance criteria, and scope boundaries. * Orchestrate specialized subagents (backend, frontend, testing, and review) coordinated by a primary agent responsible for planning and integration. * Generate and execute AI-powered tests (unit, integration, and end-to-end), verifying that the agent actually runs and fixes failures. * Critically review generated code to detect hallucinations, invented dependencies, and regressions before merging. * Build and maintain reusable Skills (SKILL.md) and MCP servers that provide agents secure access to APIs, data, and internal services. * Maintain version-controlled prompts, agent configurations, and project memory (CLAUDE.md / AGENTS.md) in the repository. * Keep the final diff small, readable, and auditable—regardless of how much code the agent generates. * Other duties as assigned by the direct supervisor. Salary: Starting from S/.4,000.00 per month Work Location: Hybrid in La Molina, Lima