Chaima Mehri

• Developed a Large Language Model (LLM) search system for construction software, ”Averroes,” aimed at enhancing user experience and boosting productivity. • Utilized innovative technologies such as Vertex AI Agents and Gemini models to refine search and retrieval capabilities. • Integrated PDF capabilities into the LLM chat system, allowing users to seamlessly upload, view, and manipulate construction documents within the software interface.

• Designed and implemented the front-end architecture for the company’s new studio. The studio enables developers to create interactive video solutions accessible through AI technology. • Improved the display interface for a Norwegian channel TV platform, ensuring compatibility and optimal viewing across various devices, including mobile phones, tablets, and desktops

• Engineered pipelines to preprocess large datasets, encoding them with CLIP model, and implementing indexing and nearest neighbor search algorithms, resulting in the development of an efficient neural search engine. • Implemented the neural search engine within the construction software, ”Averroes,” seamlessly integrating it through a user-friendly interface.

• Developed an advanced chatbot for recruiting purposes using TensorFlow and NLP techniques. • Preprocessed datasets to train and fine-tune the chatbot’s intents and entities, ensuring optimal performance, accuracy, and alignment with user expectations. • Trained and evaluated a SpaCy Named Entity Recognition (NER) model achieving an accuracy rate of 95%.