Arad Ashrafi

Technical Lead @Duologue Systems

Edmonton, AB, CA

Leading the development and optimization of multiple AI-powered communication systems, including a reintegration support service and a healthcare standards advisory platform, focusing on improving accessibility and user experienceDesigned and implemented a variety of RAG (Retrieval Augmented Generation) systems along with multi-agent architectures to handle resource retrieval, misinformation checks, and location-based recommendations. achieving enhanced response accuracy and context-awareness in AI interactions.Engineered and deployed serverless AI solutions using RunPod and cloud infrastructure, implementing error handling and failover mechanisms for production reliability. Optimized our serverless runtime for large models with 70B parameters. Fine-tuned large language models (including Llama 3.1 70B) for specialized use cases, incorporating domain-specific knowledge and maintaining high accuracy while optimizing for computational efficiency. Implemented Parameter Efficient Fine-Tuning (PEFT) techniques like LoRA and QLoRA to optimize large language models while maintaining performance and reducing computational requirements.Built and maintained backend repository with robust database solutions for managing conversation histories and user interactions, ensuring data integrity and efficient retrieval. Built and maintained RESTful APIs for integration with various services.Built React-based UIs that transformed AI workflows into engaging, insight-driven interfaces, empowering clients to monitor user analytics, manage engagement, and take actionable decisions directly within the system.Established monitoring systems for AI model performance and user interactions, enabling data-driven improvements to response quality and user experience. Created comprehensive data preprocessing pipelines for AI model training, including synthetic data generation and validation processes.