Istanbul, Türkiye
Berk Karabıllıoğlu is a passionate and innovative AI Executive with a strong background in data science, machine learning, and artificial intelligence. Currently driving the development and deployment of advanced AI solutions at Burgan Bank, Berk specializes in Large Language Models (LLMs) and chatbot projects, Leads LLM evaluation and AI ethics initiatives by designing and managing automated testing pipelines that assess model reliability, contextual relevance, and ethical alignment. Oversees the implementation of prompt guard and safety mechanisms to ensure secure and responsible AI interactions. Utilizes advanced evaluation frameworks such as GEVal and RAGAS to perform detailed performance benchmarking, supporting continuous model improvement and informed decision-making. With over five years of experience as a Team Lead at Team LunAl, Berk has led research and development projects in data science, machine learning, and AI. His tenure at Garanti BBVA Yatırım as a Senior Data Scientist and Data Scientist further honed his expertise in data manipulation, machine learning pipelines, and customer performance analysis. Berk holds a Master of Science in Data Science from Yeditepe University, where he achieved a GPA of 3.85. His thesis work focused on AI ethics and Stable Diffusion filtering layers, reflecting his commitment to ethical AI practices and cutting-edge research. He also holds dual bachelor's degrees in Mechatronics, Robotics, and Automation Engineering from Manisa Celal Bayar University and Management Information Systems from Anadolu University. His academic journey has equipped him with a robust foundation in deep learning, image processing, database management, and AI ethics. Beyond his professional endeavors, Berk is an avid enthusiast of IoT technologies, creating smart home systems powered by AI. He enjoys hosting technology podcasts, is a dedicated gamer, and actively engages in content creation. Berk's diverse interests and continuous pursuit of knowledge keep him at the forefront of technological innovation.
Driving the development and deployment of advanced AI solutions, with a primary focus on AI and LLM chatbot projects. Actively involved in LLM evaluation and oversight tasks, including model optimization, monitoring, and deployment to ensure seamless operations and scalability. Leading efforts in model feature development and implementing prompt guard mechanisms to enhance reliability, safety, and system integrity. Facilitating collaboration between business and IT units by translating business requirements into actionable tasks and overseeing their execution. Managing team performance, optimizing workflows, and contributing to strategic discussions to align AI initiatives with organizational objectives. LLM Evaluation: To ensure robust and objective evaluation of LLM performance, I have developed a custom assessment pipeline that integrates two state of the art frameworks: GEVal and RAGAS. This architecture allows for automated and repeatable benchmarking of LLM outputs across diverse scenarios. Within the GEVal framework, we utilize ground-truth-based metrics such as faithfulness, helpfulness, relevance, and coherence providing granular insights into how well model responses align with predefined expectations. In parallel, the RAGAS framework is employed to evaluate retrieval-augmented generation (RAG) workflows, focusing on metrics like answer correctness, context precision, context recall, and groundedness. By combining these complementary evaluation approaches, the pipeline enables end-to-end quality tracking from knowledge retrieval to final response generation. The resulting quantifiable scores are used not only for model comparison and selection but also to guide iterative fine-tuning and alignment strategies. This systematic evaluation infrastructure ensures the reliability, integrity, and contextual fidelity of our LLM-based applications.
We are a team that RD & Develops and solves projects in the fields of Data Science,Machine Learning , Deep Learning and Artificial Intelligence(AI)
Data Manipulation and creates ml pipeline with Python Created NPS performance analysis with Customers comments. Created Customer problem classification Created Social media analysis for competitive firms Creating DWH Data Modeling Creating Batch Learning Churn System Creating AI Modeling for tuning commission/fee models Creating specific segmentation with Clustering CRM Analytics/Data Science(Segmentation,Cohort Analyses, Churn Prediction, Customer Life Time value And prediction for investment Customers. Forcasting, Creating Reports for KPI) Machine Learning & AI (Creating Models & CNN's with Hyperparams tuning(Optuna,Griding)) Creating campaign classification for our segment customers Creating the dataset for modelling with PL-SQL,T-SQL
TEKNOFEST 2021 Artificial Intelligence in Transportation Finalist Contestant with my start-up Team LunAI Preliminary design report - 87 Critical design report - 90 Final:16/175