AI Technical Lead — Requirements, Codex Prompts, Code Audit & Production Finalization

[ABOUT THE ROLE] We use OpenAI Codex (GPT-5.6 Sol Ultra) to build internal systems, automation tools, and data products. We are not looking for a traditional developer to write everything from scratch. We will explain our business goals and what we want to achieve. Your role is to: • Understand our goals • Translate them into clear and technically sound requirements • Create implementation prompts and task instructions for Codex After Codex produces the initial implementation, you will audit the generated code and identify: • Bugs and missing logic • Architectural issues • Data inconsistencies • Security risks • Operational problems You will then help correct those issues and bring the project to a production-ready state. [KEY RESPONSIBILITIES] • Understand our business objectives, workflows, and operational challenges • Ask the necessary questions to clarify incomplete requirements • Define the scope, features, data structures, edge cases, and error handling • Create technical requirements, non-functional requirements, and acceptance criteria • Design the system architecture and implementation approach • Create detailed implementation prompts and task instructions for Codex • Audit AI-generated and existing code • Identify bugs, omissions, inconsistencies, abnormal data, and security risks • Independently verify AI output against the actual business requirements • Test, fix, deploy, and validate the system in production • Create operational and handover documentation • Take ownership of technical completeness and production readiness within your scope We do not want someone who only reports problems. We need someone who takes ownership of resolving them and ensures that the final system works correctly. [EXPECTED WORKFLOW] 1. We explain the business objective and desired outcome. 2. You ask questions and clarify the requirements. 3. You prepare the requirements, architecture, acceptance criteria, and Codex prompts. 4. We use Codex to create the initial implementation. 5. You audit the code, data, and system behavior. 6. You or Codex implement the required corrections. 7. You complete testing and confirm that the system is production-ready. [EXAMPLE OF THE INITIAL PROJECT] Our first project may involve the following advertising-data architecture: TikTok Marketing API ↓ Scheduled data retrieval Cloud Scheduler → Cloud Run ↓ BigQuery ↓ Unified Meta and TikTok Dashboard Official TikTok MCP ↓ Analysis, Q&A, and future ad management through Codex [REQUIRED SKILLS] • Professional experience in requirements definition, system architecture, and code review • Experience using AI coding tools such as Codex or Claude Code • Strong understanding of APIs, databases, and cloud infrastructure • Ability to identify errors and missing logic in AI-generated code • Ability to understand business requirements, not just technical requirements • Ability to convert ambiguous ideas into executable specifications • Willingness to take ownership of the final quality of your assigned scope • Clear and proactive communication Experience with the following is highly desirable: • GCP • Cloud Run • Cloud Scheduler • BigQuery • Marketing APIs • Next.js • TypeScript • Python [EXPECTED DELIVERABLES] • Requirements specification • System architecture and data-flow documentation • Codex implementation prompt package • Code-audit findings and implemented fixes • Test and validation results • Production-ready code • Operations and handover documentation [CONTRACT STRUCTURE] We will begin with a small paid trial to evaluate fit and capability. If successful, we would like to offer ongoing work across multiple projects. [APPLICATION QUESTIONS] 1. Please describe your experience auditing and correcting AI-generated code. 2. Please share an example where you converted an unclear business request into a complete technical specification. 3. How do you validate that AI-generated code is correct and production-ready? 4. What experience do you have with GCP, BigQuery, and API integrations? 5. Please share a previous project that is similar to this role. [IMPORTANT — READ BEFORE APPLYING] To confirm that you have read the full job description, begin your proposal with the exact phrase: CODEX AUDIT Then identify one specific failure risk in the following pipeline and briefly explain how you would test or prevent it: TikTok Marketing API → Cloud Run → BigQuery Proposals that do not include both will not be reviewed.