Enterprise Data Classification Expert to Extend Microsoft Presidio at Speed and Scale

Enterprise Data Classification Expert to Extend Microsoft Presidio at Speed and Scale Fixed Price: $1,500 We use Microsoft Presidio within an existing enterprise sensitive-data classification solution. We need a data classification expert to improve its accuracy, throughput, scalability, and coverage across large enterprise data environments. The priority is not general Python development. The priority is designing and implementing a classification system that processes high data volumes efficiently, identifies meaningful enterprise-sensitive information, minimizes false positives, and produces actionable classifications. Scope * Assess the current Presidio classification architecture, performance, entity coverage, and results. * Identify accuracy, throughput, latency, and scalability constraints. * Recommend an enterprise-sensitive data taxonomy that extends beyond standard PII. * Build and integrate five high-value custom classifications. * Improve classification precision and recall through patterns, context, validation, confidence scoring, and other appropriate detection methods. * Optimize processing for large document collections and high-volume classification workflows. * Recommend batching, parallelization, caching, model-loading, and deployment improvements. * Establish performance benchmarks for throughput, latency, and resource consumption. * Create repeatable testing for classification accuracy and performance. * Document how the classification library can be expanded without redesigning the system. Target Classification Areas * Legal and contractual information * Financial and accounting information * HR and workforce data * Security and infrastructure information * Intellectual property and confidential business information * Organization-specific identifiers, terminology, and sensitive concepts Final Deliverables 1. Current-state classification and performance assessment. 2. Recommended enterprise-sensitive data taxonomy. 3. Five integrated custom classifications. 4. Improved configuration for context, confidence, validation, and classification thresholds. 5. Before-and-after precision, recall, throughput, and latency measurements. 6. Automated accuracy and performance tests. 7. Prioritized architecture recommendations for scaling classification across large enterprise datasets. 8. Documentation and technical handoff. 9. All completed source code and work product committed to our repository. Required Experience * Enterprise data classification and sensitive-data discovery * Microsoft Presidio or a comparable detection framework * High-volume document or data-processing architectures * Classification taxonomy and policy design * Precision, recall, confidence scoring, and false-positive reduction * Pattern-based, contextual, dictionary-based, and model-based classification * Performance benchmarking and pipeline optimization * Secure enterprise data handling Strong preference for experience with: * Microsoft Purview or other enterprise data-governance platforms * SharePoint, file shares, cloud storage, or large document repositories * Parallel and distributed processing * NLP and named-entity recognition * Azure, containers, APIs, and production classification services Application Questions 1. Describe an enterprise data-classification system you designed or improved. 2. What data volumes did it process, and what throughput did it achieve? 3. How did you balance detection accuracy with processing speed and infrastructure cost? 4. How would you extend Presidio beyond standard PII into legal, financial, security, and confidential business classifications? 5. How do you measure and reduce false positives without missing high-risk information? 6. What architectural changes most improve Presidio throughput at enterprise scale? 7. Provide an example of relevant classification, detection, or high-volume processing work. 8. Confirm that you will complete the full scope for a fixed total price of $1,500. Acceptance Criteria * Five custom classifications operate within the existing workflow. * Each classification includes positive, negative, and ambiguous test cases. * Precision, recall, throughput, and latency are measured and documented. * The implementation demonstrates a measurable improvement in classification quality or processing performance. * The recommended scaling architecture is specific and executable. * Configuration and technical documentation are complete. * All source code and work product are delivered to us. Budget and Milestones The total project budget is $1,500 fixed price, including assessment, implementation, testing, performance benchmarking, documentation, revisions, and handoff. Milestone 1: Classification, architecture, and performance assessment, $300 Milestone 2: Five classifications and processing improvements, $800 Milestone 3: Benchmarks, testing, documentation, revisions, and handoff, $400 The engagement focuses on extending and optimizing the existing classification capability. It does not include rebuilding the application, creating a new user interface, or training a custom foundation model.