Hi,
I'm David
Machine Learning
Engineer

Contact

Welcome

A Machine Learning Engineer with intensive skills in Machine Learning and Deep Learning. With more than 3 years of experience on different projects that use ML and DL to solve real-world problems.
My daily job is to build Machine Learning pipelines, models and automatization of various processes. Also researching, designing, implementing and deploying scalable machine learning solutions. In addition, I usually spend my personal time on personal projects related to ML.

Frequently used

Experience

assets/imgs/epo_logo.jpg
European Patent Office, Germany
Data science
SEP 2021 - Present
  • • Designed and developed scalable ETL pipeline by using Spark, Airflow, AWS S3 handling more than 100 .million records
  • • Developed models to improve existing one giving more flexibility in the final tool.
  • • Created dashboards and reports using Tableau, Python that communicate a story and provide visualization of data in a way that can be best utilized by internal customers.
  • • Automation of regular reports saving working hours.
assets/imgs/upv_logojpg
Polytechnic University of Valencia, Spain
Machine Learning Engineer
Nov 2019 - Sep 2020
  • • Focused on projects of Deep Learning, mainly in image and generation content
  • • Creation of models using Pytorch, Tensorflow, and tracking the outputs and hyperparameters using Wandb, decreasing times between iterations of projects.
  • • Implementation of a reinforcement learning environment.
  • • Collaboration with the european commission for the definition of TRL in content generation and for a study on adversarial images and tabular data.
  • • Two papers publised.
assets/imgs/beiota_logo.jpg
Beiota, Spain
Co-founder
Sep 2018 - Jan 2021
  • • Use of technology serverless in Azure to collect data from MongoDB to table storage reducing storage cost and increasing the speed of response to the customer.
  • • Generation of report in excel and PDF to the customer using Azure Queue storage.
  • • Handled Azure Devops to have continuos integration with the infraestructure.
  • • Support in defining mock-ups. In addition, test of the UI of the platform.

Personal projects

SWISS STARTUP SCRAPPING

The project consists of carrying out web scraping the page of swiss tech After that, data has been analyzed

See more

GRAPHS FROM SCRATCH

The project consists of create a class to generate graphs and also some test

See more

AENA SCRAPPING

The project consists of carrying out web scraping of the aena page After that, data has been analyzed, preprocessing and apply NLP and GNN

See more

Streamlit site with challenges

App using the functionality provided by Streamlit. In this case my girlfriend need to solve a few riddles and if she does it will recieve a key secret to insert in the bottom. The end product is a voucher in PDF with the gift to her birthday

See more
DeepFake project

This a project that I appreciate because it was used to generate a gift to my girlfriend. Using DeepFaceLab a clip of Harry Potter was made changing harry potter's face for hers.. After this I decide use face recognition to get all the frames where Harry potter appeared in a specific movie and with this information automate the change of face in the first movie

ATP Beating

Starting from a branch this was a project for the big data exploitation subject. The idea was using data from tennis matches try to predict who is the winning in each match. And with this information decide which was the best strategy of odds comparing with real values in betting house

See more

Hackatons

Cajamar Water Footprint

Year 2022

Time series problem, without any extra feature. The target was to predict the consumption of water. To solve this problem data preprocessing was done, in addition new features was created. You can find here the code


Pawpularity Contest

Year 2021

In this competition, it was analyzed raw images and metadata to predict the “Pawpularity” of pet photos. In this project Docker and Pytorch were used to avoid any problem in relation with the environment between my colleague and me. You can find here the code


Ocean Challenge

Year 2021
Position: 2th

The competition consisted of problem of a computer vision, classification of seabed. To solve this problem was used the framework Pytorch Lightning. In addition, the library Timm was used to download pretrained models and apply transfer learning. You can find here the code

Cajamar Atmira Stock Prediction

Year 2021

Time series problem, where the target was to predict the demand of a e-commerce. Moreover in this problem you have more features that only the target. To solve this problem data preprocessing was done, in addition new features was created. You can find here the code


House price prediction subject: big data exploitation

Year 2019
Position: 1th

This contest was promoted by the professor of the subject big data exploitation between the alumns. It is a regression problem where the study of the dataset, preprocessing and the preprocessing and the testing of the different models through cross validation was essencial to win.


Education

Polytechnic University of Valencia, Spain
Master of Data Science
Degree Classification: Excelent
Note: 9/10.00
SEP 2019 - SEP 2021

Top 5 percentile of the class

    Machine Learning Specialization
    Certification: in here

    In addition to this certification I have done many more (>30) in relation AI or programming skill, so I recommend you look at my linkedin for the rest

      GET IN TOUCH

      Discuss a project or just want to say hi? Write me on Linkedin my Inbox is always open