Diya L.

• Developed and deployed customized AI solutions, collaborating with companies to define business objectives and integrate AI models, resulting in optimized performance and cost efficiency. • Organized and conducted practical AI training bootcamps tailored to industry needs, including fundamental courses, sector specific applications, and post-training support, enhancing participants' AI skills. • Evaluated AI project feasibility by analyzing strategic objectives, technological resources, and risks, ensuring alignment with company needs and successful implementation.

At Software club I was tasked to integrate AI from scratch: • Analyzing the company from both customer and team perspectives, I utilized data mining techniques to study the distribution of the population. Used models based on embeddings techniques and clustering algorithms, I was able to understand the business also producing a recommendation system for the users' platform. •Building an end-to-end complex hierarchical classification system from feasibility study to production. During the piloting phase, I engineered the evaluation metric using the previous recommendation model which helped measure and optimize performance reaching state-of-the-art results . Finally, deploying this model in production, enabling vertical and horizontal scaling by covering more market segments and companies. • Managing an industrial AI project that aims to automate a complex task, by utilizing the Standard AI Product Development Framework: I integrated AI into a new company's screening process, reducing time by 80%.

Produce an end-to-end code for the Standardization of Alstom product component descriptions using the state of the art Natural Language Processing technics.(LLMs)

As a Machine Learning Research Intern at MYTEAM.ai, I led a project to train and deploy an AI estimator based on Dynamic Time Warping and supervised machine learning to predict emotions in children. Our team's goal was to establish a more natural interaction between children and robots by analyzing their emotions through the way they perform their daily movements online. We were able to enhance the research paper algorithm to capture the right pattern using different computer vision techniques and reduce deployment time to 2 seconds per video, making our model competitive with real-time advanced deep learning models. It was an exciting and rewarding experience to contribute to the development of technology that can improve human-robot interactions.

-Data Preparation, Feature Engineering and Predictions. -Analysing fashion data from multiple seasons in order to detect market insights and predict future trends.