Hi, my name is
Mario Parreño.
I make computers learn.
I’m a Machine Learning engineer specializing in Computer Vision and Generative AI, automating problems that save companies time and money. Currently, I’m focused on tackling real-world problems with Deep Learning.
About Me
Hi there! My name is Mario, and I am a Machine Learning engineer. I specialize in Computer Vision, using neural networks to solve complex problems. I've won several national and international competitions, which have helped me to apply and understand theoretical concepts.
Through my work, I've sought to bring the abstract concepts of neural networks to life by creating tangible projects that showcase their capabilities.
You can find information about my background, skills, competition wins, and work experience on this website. You can also browse my portfolio of work, including some of my most successful projects in Computer Vision. I hope you enjoy exploring my website!
Here are a few technologies I’ve been working with recently:
- Pytorch
- Computer Vision
- Generative AI
- OpenCV
- Weights & Biases
- Team Lead
Where I’ve Worked
Sr. Data Scientist @ Cognizant
September 2022 - Present
- Led the Labelling team for a project automating processes for La Caixa banking foundation, managing 20 labelers using Google Vertex AI. Classification, detections, and text extraction problems. Continuous improvement by carrying out error analysis by expert labeling over different subsets and correction of the annotation guides for specific cases
- MLOPs implementation of a batch pipeline and an inference pipeline for a developed document management solution. Multi-Stage pipelines on Azure ML platform
- Led a project for Keysight for the detection of components in mobile applications. Computer Vision problem using detection models and OCR. Constraints on the solution as it had to run on consumer hardware and prevent data leakage
- Internal Generative AI PoCs based on LLMs: Reasons-to-buy analytics, conversational and tool-based agents
- Leading a project for Santander Bank in a Generative AI project. Extraction of fields of interest from scanned documents with different qualities. Hybrid solution based on computer vision and natural language processing applying detection models, OCR, and LLMs
Some Things I’ve Built
Featured Project
DASeGAN
Domain Adaptation and Generalization for Medical Segmentation Tasks via Generative Adversarial Networks. Placing fourth at the MICCAI 2020 M&Ms Challenge.
I developed a framework for domain adaptation and generalization using GANs to map images into a universal domain for segmentation tasks, unifying appearance across domains through adversarial training.
- Pytorch
- Weights & Biases
- Streamlit
Featured Project
Are you in the Prado museum?
Through web scrapping techniques, it is possible to collect images of the works in the Prado museum. We have hundreds of algorithms to detect faces. We can obtain embeddings that encode our faces using Siamese neural networks. We just need to get the closest embedding!
- Pytorch
- Face recognition
- ChromaDB
Featured Project
Agro Analysis
Analysis of the behavior of the Spanish fruit and vegetable market during the pandemic period with data from different sources. National competition obtaining the third place.
- Pytorch
- Pandas
- FastAPI
- JavaScript
- Docker
Other Noteworthy Projects
view the archiveSIMEPU
Automatic identification and classification of urban pavement distresses. Placed first of 62 at JAFC International Award for Innovation in Roads - VIII Edition.
Ensemble & Calibration
Study the model ensemble's effect on the calibration of the results. Tested temperature and matrix scaling over ensembles, obtaining better calibration results.
Cardiac M&Ms
Building generalizable models that can be applied consistently across clinical centers. Placed fourth of 14 researchers.
PAIP Colorectal Cancer
Automated classification of molecular subtypes in colorectal cancer for whole-slide image analyses.
Bengali Classification
Bengali.AI Handwritten Grapheme Classification Competition at Kaggle. Placed 33 of 2059 (top 2%) competitors.
Melanoma Detection
Classify dermoscopic images among nine different diagnostic categories. Placed third of 64 at ISIC 2019.
What’s Next?
Get In Touch
Although I’m not currently looking for any new opportunities, my inbox is always open. Whether you have a question or just want to say hi, I’ll try my best to get back to you!
Say Hello