Greater San Sebastian Area
PhD in artificial intelligence with experience in developing and deploying both probabilistic and deep learning models. Hands-on experience designing, innovating and shipping machine learning systems, from hidden Markov models and Bayesian networks for time series modelling, analysis, concept drift detection and forecasting, to Transformers for natural language processing / larger language models and convolutional neural networks for computer vision tasks. Comfortable during all product maturation process: research, implementation (Python / Pytorch / openCV / pandas / Numpy / fastapi / sqlalchemy / django, C++, emscripten), containerisation with Docker / Knative, control version with git, export with ONNX, and deployment on browsers for continuous time inference with small latency and small memory usage. I also enjoy other topics such as functional and numerical analysis, ordinary and partial differential equations and topological optimization.
- Developed WASM and ONNX binaries to run natural language processing and computer vision algorithm on browsers (Pytorch + Cpp + JS) - Designed a Transformer-based pipeline for spam detection to be deployed on browsers with an accuracy of 98% for both english and spanish messages - Created a CNN model for image compression in videocalls, reducing bandwidth consumption up to 75%
- Helped in the automatization of report generation regarding failures in wind turbines - Implemented algorithms based on CNN and computer vision to detect and label failures in wind turbines - Create docker images for production environments, containing APIs to use HMMs to predict drone position, and camera models to stitch photos taken by drones
- Proposed HMMs to predict the remaining useful life of industrial assets and predict industrial time series - Did code optimization and evaluation in collaboration with the Barcelona Supercomputer Center - Wrote and published research articles in high-impact journals (Q1 Journals) - Performed data analysis and report generation for clients regarding their mechanical assets using Bayesian networks
- Generated novel machine learning tools to analyze time series upon client needs - Wrote and published research articles for conferences and a book chapter
- Designed mechanical tools using topological optimization - Developed and tested digital filters to detect failures in mechanical assets - Simulated mechanical processes using finite-element-analysis to enhance clients products