Python Engineer to Extend a Document Processing Pipeline

We are looking for an experienced Python engineer to extend an existing document processing application so that it can accept Word, HTML, Excel, and PowerPoint files in addition to PDFs. We have an operational application that processes sets of publications relating to research studies. Our application currently: Reads grouped PDF files from a SharePoint folder or local folder Retrieves version-controlled extraction prompts from Supabase Submits each study’s publication set to the Claude API Stores raw LLM outputs for audit and traceability Parses LLM outputs into structured fields Populates an Excel-based data extraction table This project is focused on extending its input capabilities while preserving the current downstream workflow, audit trail, and output structure. The selected engineer will: Review the existing Python codebase and workflow Recommend an appropriate architecture for supporting multiple document formats Implement reliable ingestion and parsing for Word, HTML, Excel and Powerpoint files Integrate the new input formats into the existing study-level processing workflow Ensure files from SharePoint and local folders are handled consistently Preserve source metadata and raw Claude API responses for audit purposes Plan for handling of unsupported, corrupted, encrypted, or partially readable files Confirm that existing PDF processing continues to work without regression Document the implementation, dependencies, limitations, and configuration requirements Provide a clear handover of the completed work We are open to the engineer recommending whether documents should be converted into a common intermediate representation or processed through format-specific adapters. The proposed approach must be maintainable and must preserve the content required for accurate extraction. Deliverables A short technical assessment of the existing architecture An agreed implementation plan covering supported formats and parsing approach Production-ready Word, HTML, Excel and PowerPoint support Integration with the existing Claude API and Supabase workflow Test results for a representative set of documents Error handling and logging for failed or unsupported inputs Updated technical documentation Handover of the completed and tested code The work will be considered complete when: Supported files can enter the same study-level workflow currently used for PDFs Relevant textual and tabular content is passed through the pipeline without material loss The application retains sufficient source metadata for audit and troubleshooting Raw Claude API outputs continue to be stored Structured outputs continue to populate the existing Excel extraction table Errors are logged clearly without causing unrelated studies to fail Required experience Strong Python development experience Experience maintaining and extending an existing codebase Practical experience with the Anthropic Claude API Experience with Supabase, PostgreSQL, or a comparable backend platform Experience parsing and normalising documents in Python Experience working with structured and semi-structured content Strong automated testing and debugging practices Familiarity with Git-based development workflows Ability to explain technical decisions and trade-offs clearly Please include in your proposal: A brief description of the most relevant pipeline you have built or maintained Your experience with the Claude API or other LLM APIs Your initial view on whether to use a common intermediate representation or format-specific adapters Your availability and an indicative delivery timeline Any access or technical information you would need before providing a detailed estimate Please focus your proposal on directly relevant experience rather than sending a general development profile.