We are looking for a specialist who has practical experience in building or refactoring complex development processes using AI agents (reference - Claude Code Agent Teams / LangGraph-like solutions) for a large multi-tenant SaaS.
What needs to be done
• Design and configure the orchestration of AI agents for our stack:• Backend: PHP 8.1 + Yii2
• Frontend: Vue 3.4 + Vite (6 SPA)
• DB: MySQL / MariaDB
• Redis, Docker, Nginx, GitLab CI, Codeception/PHPUnit, Vitest/Playwright
• Build a team of 10–14 AI agents (orchestrator, backend, frontend, QA, DevOps, Bug Tracker, Integration, Designer, Mobile, Chatbot, BA/Architect, Documentation, Security, etc.) with clear roles and rules:
• Autonomously: tests, migrations, documentation, new modules
• Only with human audit: business logic, routing, finance
• Prohibited: prod-deploy, secrets, RBAC
• Set up a full development cycle through agents:
from task in Trello/GitHub → decomposition → development → tests → Code Review → PR.
• Build a project knowledge base (architecture, business rules, integrations, security) and connect it to AI processes.
• Train the team: how to assign tasks to agents, how to review PRs, how to work with a hybrid autonomy model.
Expected experience
• Real practical experience in creating or optimizing:
• AI orchestration (agent teams, multi-agent systems),
• development pipelines with AI,
• or complex technical systems with deep integration into CI/CD.
• Understanding of SaaS architecture, multi-tenancy, security (RBAC, secrets, audit log).
• Experience with VS Code / Cursor, GitHub/GitLab, AI tools for coding.
• Ability to explain, teach, and document processes for the development team.
Initial tasks
• Propose the optimal architecture of AI processes for our stack and scale.
• Estimate the scope of work, phases (phase 1–3), and collaboration model.
• Demonstrate on one real task (for example, “Loyalty System”) the full cycle: from task assignment to completed PRs.
Collaboration format
• Freelance, phased payment (phase 1–3, with interim acceptance).
• Estimated timeline: 3–4 weeks for full implementation + support period.
• Mandatory NDA.
If you have successful cases of implementing AI agents in real product teams — briefly describe them and add links (repositories, articles, demos).