Ozlem Akpinar Ciaccio

• Worked on Kore.ai’s Conversational AI chatbot for Belfius Bank • Improved response quality and intent accuracy with precise language adjustments in daily testing

* Develop natural language understanding technology (NLU) and advanced conversational dialog system to produce custom domains for in-car voice assistants for major car makers, including Audi, BMW, and Mercedes. * Curating linguistic datasets to refine NLU model training * Developing lexicons and ontologies to bolster semantic understandingImplementing linguistic features to enhance NLU accuracy * Training of embedded and cloud machine learning models for NLU * Establishing guidelines for entity tagging and categorisation * Training models on annotated datasets for enhanced language comprehension * Optimized existing models to improve performance and accuracy for embedded and cloud solutions to reach at least 90% different domains. * Managed weekly language model deployments using an existing Jenkins pipeline.

• Manipulating and managing large amounts of datasets, Excel, and CSV files • Root cause analysis and uncovering issues using data for different domains, then assisting the model improvement • Annotating data and finding appropriately annotated datasets for Supervised Learning methods • Working with QA testers and UX team to optimize contexts development for Bixby applications • Interpreting data, analysing results, and providing ongoing reports to improve language models • Analysing large, complex data sets representing the behaviour of users to optimise the quality of the training data • Handle, track, and measure quality from manual data validation

Statistical Language Modeling for Turkish Language