AI Microservice for Document Processing Platform (CPU-based) Project Overview We are developing a document management platform that processes different types of documents such as administrative letters, scanned files, and business documents. We are looking for an experienced AI / NLP developer or team to build a lightweight AI microservice that will assist users in analyzing documents and extracting useful information. The AI must act as a support tool, meaning it only suggests information, while users always validate the results. Main Features (V1) The AI service should provide the following capabilities. 1. Document Summarization Generate a short summary (50 to 150 words ) according to the from the text content of a document. Typical workflow: 1. Text is extracted from a document (OCR by the platform). 2. The text is sent to the AI service. 3. The AI returns a concise summary. 2. Metadata Extraction The AI should extract key metadata from documents. Examples: • subject • sender • recipient • document date • reference number • document type • keywords • .... 3. Completion of Missing Metadata The AI service should analyze document text and suggest values for missing metadata fields when documents are imported or scanned. The AI should only run when: • metadata fields are missing • documents come from external sources • a user explicitly triggers the AI Existing metadata must never be overwritten automatically. Architecture The AI must be implemented as a separate microservice. Simplified architecture: User ↓ Frontend ↓ Backend API ↓ AI Service Communication with the AI service will be done through REST API. Example endpoints: POST /ai/summarize POST /ai/extract-metadata Technical Requirements • Have to be run on CPU infrastructure (no GPU required) • Must support on-premise deployment • Prefer Python • NLP / LLM-based text processing • Asynchronous processing • Docker deployment Deliverables The developer or team should deliver: • AI microservice • REST API endpoints • documentation (API specification) • Docker deployment • integration guidelines for the backend team