PostgreSQL Developer — Migrate 4D Business Logic to PostgreSQL

We are migrating a set of data-processing routines from a legacy 4D (with MySQL) desktop-database system to PostgreSQL running on a Linux server. We're looking for an experienced PostgreSQL developer to lead this re-implementation and validate it against the existing system's output. Importantly, this is an AI-assisted project. The source system is written in 4D, which is a very niche technology, and we do not expect you to already know it. We use AI tooling (Claude) to read and explain the original logic and to draft the PostgreSQL equivalents. Your job is to drive that process, review and correct the AI's output, test it, and get it working correctly — not to be a 4D expert. This is a scoped, well-defined project with clear deliverables and a structured review-and-sign-off process. There is a realistic path to further work for the right person, but this posting covers one specific piece. What the work involves The existing system runs a daily batch process: it imports raw input files, performs a series of calculations and data-quality checks, and produces a set of output tables that feed downstream reporting. Your task is to reproduce this processing in PostgreSQL so that it generates the same outputs from the same inputs — then measure and document the performance improvement. In practice the workflow is: use AI (Claude) to explain each original routine and generate a first-draft PostgreSQL version, then apply your own expertise to review, correct, test, and validate it against the known output. The AI accelerates the work, but a skilled human is essential — its output is a starting point, never the finished product. The work is broken into sequential stages (import, commit/derive, validation, core calculations, trend/aggregate calculations, and exports). Each stage is completed, tested, and signed off before the next begins. Much of the logic follows repeating patterns, so once an approach is agreed for one routine, similar ones can be templated. Key requirements Strong PostgreSQL skills: complex queries, CTEs, window functions, aggregation, and set-based rewrites of procedural logic Comfortable working alongside AI coding assistants — prompting them well, then critically reviewing, correcting, and testing what they produce rather than trusting it blindly Disciplined about correctness: the priority is an exact functional match to the current output first, optimization second Able to build test/validation routines that compare your results against a known reference output Clear written communication and a methodical, documented approach No prior 4D experience required — comfort reading unfamiliar logic (with AI assistance) matters far more How we work The original logic is niche and AI-assisted translation is central to how we work — we'll set you up to use Claude for reading and drafting Mirror the existing behavior exactly first; propose optimizations only once results are confirmed correct Each routine is analyzed, re-implemented, tested against reference output, and reviewed before it's marked complete Execution timings are recorded so we can quantify the performance gain All AI output must be understood, tested, and verified by you — nothing is accepted as production-ready without human review To apply Please share relevant PostgreSQL experience and a brief note on how you work with AI coding tools — specifically how you validate and correct AI-generated SQL rather than taking it at face value. Include your availability. You don't need 4D experience; tell us instead how you'd get comfortable reproducing unfamiliar logic with AI support.