Juan Fonseca Núñez, Data Scientist senior

Consulta la experiencia profesional de Juan

Engineer and Data Scientist with more than 4 years of experience developing projects to create value in companies, finding implicit patterns in data for its actionability, Machine Learning models, Deep Learning and extensive experience in its applicability to Computer Vision. Cum Laude distinction Engineer, MSc in Data Analysis, Statistics and Optimization from the Universitat Politècnica de València, and currently finishing a MSc in Artificial Intelligence at the Universitat Politècnica de Catalunya (BarcelonaTech).

Experience as Data Science consultant guiding and developing projects in different high impact areas, I have experience making predictions with time series, fraud prevention models with financial institutions, prediction of customer behavior, optimization of incentives with mathematical models of integer programming and uplifting. Later I worked as Computer Vision Engineer and Data Scientist using security cameras to obtain patterns of customers in the retail and security sector, creating and using models of people detection, facial detection and recognition and its incorporation into real-time databases for integration and exploitation.

Finally I work in the Cyber Security Operations Center Data Science team at Nestlé, researching and improving the models using Adversarial Machine Learning and Explainable AI, in order to keep the models at the forefront and ensure their robustness against attacks, such as phishing in corporate emails, imitation of logos in fraudulent pages and atypical behaviors.

I work mainly using Python using frameworks such as scikit learn for traditional ML models, boosting, bagging and unsupervised, and TensorFlow for Deep Learning models. I have extensive knowledge of R, but I am using it occasionally due to the potential of Python nowadays.

Nestlé

Data Scientist at Nestlé CyberSOC

HIGH TECH

febrero de 2021 - Hoy (3 años y 12 meses)

Barcelona, España

Adversarial Machine Learning researcher in the Cyber Security Operations Center Data Science team. Improves phishing and fake webpages detection models by improving ML algorithms with modern Adversarial Machine Learning methods. Deploys the created models and KPIs using Azure DevOps jobs.

Deep Learning NLP Boosting Machine Learning Adversarial Machine Learning TensorFlow Scikit-learn

Grupo Alto

Data Scientist & Computer Vision Engineer

HIGH TECH

noviembre de 2019 - septiembre de 2020 (10 meses)

Bogotá, Colombia

Member of the algorithms and solutions research team, working on Machine Learning, Deep Learning and Data Science.

Develops and applies algorithms and methodologies for the extraction of information from unstructured data of

CCTV videos, evidence images and text, in order to provide additional information to the company's traditional

transactional data, using Computer Vision, Deep Learning and NLP techniques.

Takes advantage of the extracted data to use data science techniques to advise and make consultancy for the clients

of the retail sector in order to reduce their operational losses, and to extract value insights about their clients.

Some of the algorithms implemented through open source tools (Python, TensorFlow, PyTorch, R):

Face detection and recognition (MTCNN and FaceNet).

Object detection and tracking (YOLO v3, DeepSort, CenterTrack)

Transfer Learning with YOLO v3 for the detection of objects of interest, in order to guarantee security and extract

data from self-checkout videos.

CNN training for gender, age, race and emotion estimation from faces, as well as for the classification of large

volumes of images.

Deep Pose implementation

OCR for automatic text extraction from multiple nationalities IDs.

Use of traditional Computer Vision techniques to detect dark glasses on faces, detect store closures, receipt printing,

and overall to maximize information extraction from images useful in retail

NLP for text-based entity analysis of court cases

Boosting techniques for the prediction of scalable multivariate time series

Clustering for signal clustering, client characterization, and face recognition analysis optimization

Deep Learning machine learning Python TensorFlow Pytorch Consultoría Data science

Rappi

Data Scientist

HIGH TECH

julio de 2019 - noviembre de 2019 (4 meses)

Bogotá, Colombia

Works in the Growth team aiming to retain and accelerate the growth of users in the application.

Develops and automates time series models for 7 countries based on ARIMA and XGBoost models to predict the

behaviour of (voluntary) organic purchases on a daily basis and calculate the incentives needed to meet period

targets.

Develops Uplift models to avoid giving incentives to users with high purchasing potential, creating a methodology

and automating campaigns for 7 countries.

Undertakes design and analysis of experiments in order to maximize the long-term incremental GMV of users

Leads the creation of Machine Learning models for the Expansion area, in which it uses data exogenous to the application

through Web Scraping, and applies models to predict GMV and user scores for each of the candidate cities by quadrants

machine learning Data science Scikit-learn SQL Statistics time series

Consulta todas las experiencias de Juan

MSc Data Analysis, Process Improvement and Decision Support

Universitat Politecnica de Valencia

2017

Statistics and Operations Research Academic grade: 9.23/10 Honours in two subjects Honours Thesis: Mixed Integer Linear Programming Models for Production Planning with short life articles (http:// hdl.handle.net/10251/89439) Data Analysis: Data Mining, Forecasting Techniques, Multivariate Analysis, Neural Networks and Simulation Process Improvement: Multivariate Process Analysis, Monitoring and Diagnosis, Advanced Linear Regression Models and ANOVA, Design of Experiments, Statistical Process Control, Reliability, Availability and Mantainability. Decision Support and Operations Research: Modeling and Optimization, Multi-criteria Programming, Production Planning and Scheduling, Project Management Complementary Courses: Customer Satisfaction Analysis, Quality Management and Improvement

MSc Artificial Intelligence

Universitat Politecnica de Catalunya

Finished Courses: Computational Intelligence, Computational Vision, Introduction to Human Language Technology, Introduction to MultiAgent Systems, Introduction to Machine Learning, Planning and Approximate Reasoning, Ethics in AI, Deep Learning, Unsupervised and Reinforcement Learning, Explainable AI

The Python Mega Course

Udemy

2020

Python

Deep Learning Specialization

Deeplearning.ai (Coursera)

2018

Deep Learning TensorFlow

Data Science Specialization

Johns Hopkins University (Coursera)

2017

R Data science

IBM SPSS Modeler Professional V3

IBM

2018

SPSS Modeler