Jorge Veiras Email & Phone Number

London, Gb

London Area, United Kingdom

London Area, United Kingdom

🚀 Leading the development and deployment of the first-ever SaaS platform dedicated to enhancing restaurant experiences through the analysis, classification, and optimisation of their vibes.📊 Compiled a dataset of over 38,000 images through web scraping and AI-generated content to support model training.🧠 Trained and deployed Vision Transformers (ViTs), achieving up to 30% greater F1 scores in some ambience categories compared to leading commercial models like ChatGPT-4o mini and Gemini 1.5 Flash.💻 Designed and implemented both the front-end and back-end of the platform, ensuring seamless integration of features such as AI-driven ambient music generation 🎶 and optimal image selection.🤝 Collaborated with industry experts at Google to refine AI model performance and platform deployment, leveraging Google Cloud Platform (GCP) for scalable, distributed computing solutions.💷 Generated £720 in revenue in August 2024, enhancing the ambience in over 36 restaurants through data-driven insights and AI tools.

Santa Cruz De Tenerife, Canary Islands, Spain

🌊 Seagrass Ecosystem Mapping and Blue Carbon Stock Assessment🛰️ Harnessed Sentinel-2 satellite imagery to map Cymodocea nodosa’s spatiotemporal distribution and estimated blue carbon stocks.🛠️ Performed rigorous image pre-processing: atmospheric correction, radiometric correction, BRDF correction, land masking, sun glint correction, deep water masking efforts, and water column correction using depth invariant indexes.🤖 Leveraged machine learning algorithms for image coverage classification. Random Forest classifiers exhibited the highest F1 scores, ranging between 0.92 and 0.99.📈 Improved metric results from prior studies in the same study area, which yielded accuracies of 0.73 using the WV-2 satellite and the same water column correction method.🌱 Established that in 2021, in our study area, the Cymodocea nodosa sequestered, on average, 73,500 Mg of CO2, offsetting the annual CO2 emissions generated by the Canary Islands’ combustion plants.

Kosice, Slovakia

🫀 Application of Deep Learning Methods in ECG Recordings for Arrhythmia Detection🧑‍💻 Developed machine learning and deep learning algorithms for classifying arrhythmias in ECG recordings extracted from the PhysioNet database.📊 Identified that the best classifiers are Convolutional Neural Networks with an AUC of 0.9.🛠️ Acquired expertise in: machine learning and deep learning algorithms, loss functions and regularisation, optimisation methods (Gradient Descent, SGD, Adam, Nadam), cross-validation, backpropagation, activation functions (Sigmoid, ReLU, TanH, Maxout), batch normalisation, data cleaning methods, PCA, and metrics for statistical analysis.

Master'S Degree, Environmental Data Science And Machine Learning

Grade: Distinction Core modules: - Numerical Programming in Python - Computational Mathematics - Data Science and Machine Learning.

Bachelor Of Science - Bs, Physics

Thesis: From space to sea. Assessing the ability of satellite data to map benthic habitats using machine learning algorithms.