Case Study
GenAI-powered HR chatbot for Pegasus Airlines
Pegasus Airlines ยท via NTT DATA Business Solutions
Aug 2022 - Jul 2025
Building a production HR chatbot with caching, rate limiting, and cost controls around real-world LLM usage.
- Java 17
- Spring Boot
- GenAI APIs
- Redis
- Rate limiting
Challenge
HR needed a conversational assistant for employees, but naive LLM integration would have meant unpredictable API costs and no protection against traffic spikes.
Approach
- Built the chatbot as Spring Boot microservices with a clean boundary between conversation logic and the LLM provider.
- Implemented Redis-based intelligent caching so repeated questions never hit the LLM twice.
- Added rate limiting and token-cost optimization strategies tuned for production LLM usage.
Outcome
A production GenAI service with controlled, predictable API costs and stable behavior under load - one of the earliest LLM systems shipped to production in the company's portfolio.