Adrià Ciurana Lanau
Verified Expert in Engineering
Computer Vision Developer
Barcelona, Spain
Toptal member since November 24, 2021
Adrià started programming at 11 with his father. He has explored many areas, but what has fascinated him the most has been IA algorithms. Adrià specialized in classical computer vision, where high level knowledge of machine learning models was necessary. With the emergence of deep learning (DL), he specialized in DL. For eight years, he has worked on disruptive innovation projects in IA and was a trainer in machine learning, deep learning, and statistics to companies and private schools.
Portfolio
Experience
- Computer Vision - 7 years
- Deep Learning - 7 years
- Machine Learning - 7 years
- Artificial Intelligence (AI) - 7 years
- Data Pipelines - 5 years
- PyTorch - 4 years
- Semantic Segmentation - 3 years
- Instance Segmentation - 3 years
Preferred Environment
Linux, PyTorch, Slack, Visual Studio Code (VS Code), Git, Discord
The most amazing...
...project I’ve collaborated on was as the main IA developer, advisor, and trainer in a company that is one of the most promising InsurTechs in the world.
Work Experience
Deep Learning Research Lead
Photoslurp
- Developed various concept-oriented proofs for the generation of synthetic images for marketing and, as a result, obtained public financing.
- Created tools for the extraction of information in images from social networks to carry out automatic tagging, captioning, and measurement of the quality of the image, and measurement of impact.
- Developed a complete pipeline of realistic generations of virtual environments for marketing.
Founder
DreamLearning.ai
- Created my own consulting firm, which allowed the realization of successful projects.
- Offered services for project management, advising, and realization of projects in the field of artificial intelligence.
- Created different free courses at conferences for the scientific dissemination of artificial intelligence.
Deep Learning Advisor | Research Lead
BDEO
- Created a core product for the estimation of vehicle damage, as well as subsequently its quantification in the product’s cost.
- Trained workers and advisors on decision-making and the strategic vision of the company.
- Won several awards, including the best algorithm of the year, and was a two-time South Summit winner. The company is currently one of the top 100 most important insurtechs in the world.
Research Lead
Media4us
- Managed a team focused on the development of a non-intrusive advertising insertion application in sports videos.
- Created neural networks for precise segmentation of games and objects on the playing field. Carried out a shadow estimation to insert the advertisement virtually.
- Created a complete pipeline to process videos in real-time.
Deep Learning Research
MediaPro
- Developed highly efficient neural networks for the precise detection and classification of players and the ball on a soccer field.
- Optimized the neural networks used to apply them in embedded systems using libraries like TensorRT and CUDA.
- Developed a complete tool for mass data annotation in videos of various sports. These included cloud management to pseudo-annotate, manually annotate, manage data in the cloud, and automatically distribute annotation tasks to different annotators.
Professional Trainer
Datahack
- Created courses for companies in the field of machine learning.
- Presented various projects to the shipping and logistics industry to solve multiple logistical problems involving techniques of machine learning, deep learning, sensors, and big data.
- Created courses for the unemployed in the field of big data and deep learning.
Deep Learning and Machine Learning Developer
SOAX
- Developed a complete facial recognition pipeline, including web verification middleware, life detection using eye-blinking, face detection and tracking in the browser, a biometric information extraction server, and identity verification.
- Created an automatic system for extracting information from a personal document. Detected documents via webcam, the automated classification of the document, and the extraction of the relevant information through OCR techniques.
- Certified delicate processes for banks, such as digital signatures and notarial procedures, among others.
Full-stack and Machine Learning Developer
Estimtrack
- Developed a fully automated operating room management platform, allowing monitoring its current status and efficient management.
- Tracked and fixed bugs using Asana as an internal task management tool and methodologies using Agile.
- The platform also includes an estimation system for intervention times and a scheduling system to automate the intervention schedule.
Intern
Computer Vision Center
- Developed neural networks to estimate the global light of an image, as well as possible directional lights.
- Developed custom layers in CUDA to integrate them into the Caffe neural network framework.
- Developed color clustering techniques to harmonize images.
Experience
Photoslurp: Image Generation
I created all the necessary tools for data annotation, the pipeline, and study of the current state-of-the-art synthetic imaging and image analysis tools to extract valuable data and automatically generate tags, captions, and relevant information from social networks.
Bdeo: Car & House Damage Estimation
During a three year period, we combined different areas from machine learning techniques to detect, segment, and estimate the cost of damages produced in vehicles. Later we combined natural language descriptions with images to produce similar estimations from homes. To carry out this project, we developed algorithms totally from scratch. We also made complex annotation tools to accurately annotate precise data. Finally, we implemented interpretation tools to guarantee excellent results. Within the collaboration, we also developed many additional topics: fraud detection systems, recommendation systems, data management, NLP pipelines, and tools based on natural language.
Ogulo: Geometric Estimation in Homes
http://www.ogulo.comI researched cutting-edge processes and specific improvements to the models to offer more robustness. In addition, a study of the state of the art technology was carried out together with the transfer of particular knowledge to the company's technical department.
Algoritmia
I worked on an advice service regarding how to approach stock estimation problems in the technical field.
Education
Master's Degree in Big Data
Datahack School - Barcelona, Spain
Master's Degree in Computer Vision
Autonomous University of Barcelona - Barcelona, Spain
Bachelor's Degree in Computer Science
Autonomous University of Barcelona - Barcelona, Spain
Certifications
Certified Self-Driving Car Engineer
Udacity
Certified Data Engineer
Udacity
Certified Big Data Architect
Datahack schoool
Certified Deep Reinforcement Learning
Udacity
Skills
Libraries/APIs
PyTorch, NumPy, Keras, TensorFlow, Scikit-learn, Pandas, Vue, SpaCy, Rasa NLU
Tools
Slack, Git, MATLAB, Apache Airflow
Languages
Python, JavaScript, HTML, PHP
Platforms
Linux, Visual Studio Code (VS Code), Amazon Web Services (AWS), Amazon EC2, Azure, Docker
Paradigms
Scrum, Management, Dynamic Programming
Storage
Redshift, Data Pipelines, MySQL, Databases, MongoDB, Elasticsearch, Amazon S3 (AWS S3)
Frameworks
Hadoop, Spark, Caffe, Darknet
Other
Machine Learning, Computer Science, Computer Vision, Computer Vision Algorithms, Deep Learning, Generative Adversarial Networks (GANs), Object Detection, Instance Segmentation, Semantic Segmentation, Classification, Data Science, Video Analysis, Artificial Intelligence (AI), Image Retrieval, Cloning, User Behavior, Optimization, Clustering, CSS5, Alpha Matting, Bayesian Neural Networks, Convex Optimization, Research, Non-differentiable Optimization, Depth Estimation, Time Series, Time Series Analysis, Fourier Analysis, Mathematical Modeling, Data Engineering, Genetic Algorithms, Discord, Cryptography, Applied Mathematics, 3D Reconstruction, Apache Cassandra, Reinforcement Learning, Deep Reinforcement Learning, Autonomous Navigation, PID Controllers, Bayesian Statistics, Big Data Architecture, Big Data, Statistics, NVIDIA TensorRT, Natural Language Processing (NLP), Multiprocessing, GPU Computing, Graph Neural Networks (GNNs), Graphs, Geometry, Projective Geometry, Ensemble Modeling, Generative Pre-trained Transformers (GPT)
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring