Mayte Guerra Saborit

Process automation saves time on repetitive and easily reproducible tasks such as sending automated emails, scheduling appointments, and sending automated messages or using an intelligent agent. I have created intelligent conversational agents using the elevenlabs platform. I have solved real-world problems by applying machine learning and linear regression techniques. I have analyzed patterns in data using statistical and unsupervised learning techniques. I have classified images using libraries such as TensorFlow and PyTorch. The agents I have developed have been for assisting customers who want to buy a car. The image classification algorithms used have been for classifying the body condition of a cow through images. I have also used algorithms such as K-means to identify data that follows the same pattern and group it to analyze its common characteristics. I have used algorithms such as ensembles, which use fuzzy logic and the handling of unbalanced databases, to predict whether or not high-power circuit breakers require maintenance. I have also used algorithms such as kNN to predict the strength capacity of connectors in composite steel and concrete structures. I have performed fine-tuning on the OpenAI platform using less expensive models such as GPT-40-Mini, as well as preparing the data so that it can be processed by vectorized and tokenized embeddings for inclusion in OpenAI vector stores or Elleven Labs knowledge bases. This allows the models to have greater context for the problem they are solving, directly oriented to the specific problem of the company using Retrieval Augmented Generation (RAG) techniques. Furthermore, I have developed APIs with FastAPI to link these models to websites developed in languages ​​other than Python.