FinTech Data Engineer: Claude AI Workflow for Form ADV & FINTRX Analytics

We are seeking an expert Senior FinTech Data Engineer / AI Developer based in Canada or the United States to architect and build a repeatable, highly resilient wealth-management research workflow.

The objective of this project is to eliminate manual research by identifying Registered Investment Advisors (RIAs) showing documented structural or operational pressures (e.g., fee complexity, platform fragmentation, or capacity constraints) that an institutional platform or a larger RIA could address. Note: Predicting or assessing whether an owner wants to move is strictly out of scope.

This requires an engineer who can blend strict programmatic Python data pipelines with non-deterministic LLM document extraction.

Required Workflow & Technical Scope

Starting with a raw FINTRX Excel export file, the automated system must execute the following pipeline:

• Automated Document Retrieval: Dynamically extract CRD identifiers from the spreadsheet, handle SEC/IAPD connection protocols safely (respecting rate limits), and fetch the corresponding Form ADV Part 1 (JSON/XML data) and Part 2A Brochures (Narrative PDFs).
• Claude-Driven Data Reconciliation: Leverage the Claude API (via structured Tool Use / Function Calling) to parse the narrative prose of Part 2A brochures, cross-reference data against Part 1 fields, and reconcile findings directly against the initial FINTRX records.
• Non-Destructive Excel Automation: Programmatically update a multi-sheet master workbook (using openpyxl, xlwings, etc.). The script must preserve original values and dynamically flag discrepancies, missing fields, or target mismatches via a clean exception logging framework.
• Screening & Scoring Engine: Implement deterministic data logic to score and rank RIAs based on core proxies for operational pressure (such as account-to-advisor bottlenecks, tech stack fragmentation, or stagnant asset growth metrics).
• Data Integrity & Logic Separation: Force the AI pipeline to rigorously isolate verified regulatory facts from logical targeting hypotheses within the generated target profiles.

Required Qualifications & Experience

• Seniority: 5+ years of experience in Python Data Engineering and Software Architecture; 2+ years of hands-on production experience orchestrating LLM APIs.
• Domain Expertise: Prior experience working with SEC EDGAR / IAPD systems, mapping Form ADV data, and interpreting wealth management data structures (specifically FINTRX).
• Advanced LLM Engineering: Deep understanding of the Claude (Anthropic) API ecosystem, including context window management, system prompt optimization, and forcing deterministic JSON outputs through structural tool use.
• Data Architecture: Elite spreadsheet automation skills. You must know how to surgically manipulate, format, and audit Excel files via code without destructive overwrites.
• Location Constraints: Must be an independent freelance contractor legally authorized to work and residing in the United States or Canada.