Azure Platform Engineer

10+ Years

Irving, TX Hybrid (3 days onsite)

Candidates has to go for Inperson Interview for the Client Round

• Design, train, and optimize AI-powered invoice and contract OCR solutions targeting 98 99%+ extraction accuracy.
• Analyze documents across suppliers, countries, languages, formats, and scan qualities; define document taxonomy and extraction schemas.
• Work with Azure AI Document Intelligence prebuilt, custom neural, template, classification, and composed models.
• Build and maintain custom document classification and extraction models, including labeled/versioned datasets.
• Implement document preprocessing for rotation, skew, noise, resolution, and page separation.
• Develop confidence scoring, validation rules, and human-in-the-loop review workflows.
• Validate critical fields such as totals, taxes, currencies, dates, PO/supplier matching, duplicates, and cross-field consistency.
• Build automated evaluation pipelines using accuracy, precision, recall, F1, false-positive/negative rates.
• Perform error analysis, model monitoring, drift detection, retraining, and model versioning.
• Collaborate with Procurement, Supply Chain, AP, Legal, and Engineering teams.

4+ years in ML, OCR, Computer Vision, NLP, or Intelligent Document Processing. Hands-on Azure AI Document Intelligence or similar enterprise IDP platform. Strong Python, SQL, REST APIs, JSON, Azure SDKs, and asynchronous processing. Experience with document classification, field extraction, tables, and variable-length line items. Strong understanding of accuracy, precision, recall, F1, confidence calibration, and test-set design. Experience with invoices, contracts, POs, or supply-chain documents.

Azure Machine Learning, MLflow, Azure AI Search, Azure OpenAI / GenAI, Databricks. Multilingual document processing and contract/legal clause extraction. ERP/procurement systems: SAP, Oracle, Dynamics 365, Coupa, Ariba. Active learning, model-drift monitoring, human-in-the-loop, privacy, and document retention.

For applications and inquiries, contact:View email address on click.appcast.io