Manual processing of travel itineraries across formats (PDFs, emails, text) was time-consuming, error-prone, and blocked real-time integration with their AXUS travel management system. The client needed a scalable, AI-powered solution to auto-extract structured data in JSON.

Developed a document processing pipeline using LLMs (Llama) to intelligently extract, validate, and standardize itinerary data. Enabled multi-format support, real-time editing, and seamless integration with APIs.

•  Built UI & API Interface for easy file uploads and edits
• Parsed PDFs, emails (.eml), and text using PyPDF2 and email parsers
• Used LLMs for context-aware data extraction into schema-compliant JSON
• Enforced data validation (type checks, field cleaning, schema rules)
• Auto-generated titles, price totals, and date ranges
• Integrated final output with AXUS via REST API
• Deployed with session persistence, smart error handling, and token tracking
• Single GenAI model on Akamai with customizable app layer.

• Efficiency – 80% less manual work, fewer errors.
• Consistency – Standardized, accurate outputs.
• Scalability – Handles multiple formats at once.
• Adoption – Easy to use, minimal training.
• Cost Savings – Faster, frees staff for higher-value work.
• Integrates with existing systems

Deployed on Akamai’s global edge platform using containerized environments with SSL/TLS, auto-scaling, and intelligent load balancing. Akamai’s edge computing supports low-latency AI inference and secure API integrations. Cached responses, real-time monitoring, DDoS protection, and WAF ensure reliability and compliance for travel data workloads.