- Co-founded a startup with an AI-based solution to detect and diagnose failures in solar power plants, aiming to reduce costs and losses due to failures.- Developed a computer vision algorithm to identify failures in solar panels from aerial thermal images.- The project is backed by USQAI, a startup incubator.Technologies Used:- Languages: Python- Frameworks: Pytorch, ML Flow, TensorFlow, Keras, NumPy, pandas, pyplot, seaborn, sklearn, Flask, OpenCV.

- Worked in a research project on AI hardware accelerators, Edge AI and nanoelectronics for energy-efficient within the ANID FONDECYT initiation project N°11230679. - Developed a speech recognition AI model using nanoelectronics for emulating the neural network directly on the hardware to reduce the energy consumption of the model. Evaluated the model by performing speech recognition on the FSDD dataset. - This approach reduced the neural network size to 10,000 trainable parameters, which is approximately 1000 times smaller than traditional architectures like MLP, CNN, or RNN, with a simplified training algorithm. This highlighted the approach's efficacy in reducing the energy consumption of AI.- Studied the implementation of a magnetic device called Spin Torque Nano Oscillator (STNO) as a nano-electronic component for processing AI workflows at low energy consumption for edge AI.- Researched state-of-the-art hardware technologies and new computational frameworks, focusing on alternatives to the von Neumann architecture and incorporating beyond CMOS electronics such as Spintronics and Neuromorphic Computing, advancing the boundaries of modern AI hardware.- Collaborated with a team of PHDs, working under their guidance to align theproject objectives with the latest advancements in nanoelectronics, neuromorphic computing and Artificial Intelligence.- Performed EDA, data cleaning, data visualization, data preprocessing and feature engineering to feed the dataset to the model, applying statistical techniques such as standardization, normalization, outliers’ analysis, among others. - Derived an analytical expression for simulating and studying the operation of the magnetic device and the hardware circuit. Studied the effect of the size of the magnetic hardware device on the performance of the AI. Technologies Used:- Languages: Python- Machine Learning Frameworks: TensorFlow, NumPy, pandas, pyplot, seaborn, SciPy, sklearn.

- Developed a full stack web page for a real estate agency, providing a comprehensive and user-friendly property catalog.- Designed and developed the user interface for an intuitive and engaging user experience.- Ensured the security and scalability of the application using AWS services.- Managed version control and code collaboration through Git.- Web page developed: https://www.ofrecearriendo.com/Technologies Used:- Frontend: HTML, CSS, JavaScript, Bootstrap- Backend: Python, Node.js. - Cloud: Git Source Control, AWS.