At Meezan Bank, I structured backend modules with Spring Boot, using object-oriented design, interfaces, and dependency injection to keep code testable and maintainable.
I also built Angular features for a production monolithic application alongside senior engineers. Redis caching reduced repeated database calls on high-traffic endpoints, and RabbitMQ workers moved email and image processing off the main thread, reducing response time from seconds to milliseconds.
At the National Center of Artificial Intelligence, I used H2O.ai to preprocess and model datasets of 5K+ records for predictive analysis. I also designed and optimized prompts for LLMs and developed a RAG chatbot that retrieved information from 100+ document pages.
In my projects, I contributed to a multi-agent simulator that reduced simulation tick latency 27x, and built a crypto analytics backend processing 100+ metrics under 300ms. I’ve also worked on bilingual speech emotion recognition, preparing English and Urdu recordings and evaluating deep learning models.
