Enrique Balp-Straffon
Verified Expert in Engineering
Data Scientist and Developer
Enrique is a data scientist with an academic background in physics and neuroscience. Over the years, he has participated and led teams in several machine learning projects with both startups and corporations in the financial, healthcare, logistics, marketing, and energy sectors. Enrique has in-depth experience in different areas of artificial intelligence, such as computer vision, natural language processing, financial risk modeling, and so on.
Portfolio
Experience
Availability
Preferred Environment
Docker, TensorFlow, Scikit-learn, NumPy, Pandas, Python
The most amazing...
...project I've made is an auto-ML platform for the automatic searching of the best data preprocessing strategies, hyper-parameters, and deep neural architectures.
Work Experience
Senior Data Scientist
Wesper
- Developed a sleep expert chatbot with LLMs and OpenAI APIs, creating a RAG vector DB-based pipeline with expert knowledge.
- Created ML algorithms to extract precise insights regarding sleep health from body sensor data, including a sleep-wake model and a respiratory event detection model.
- Achieved state-of-the-art results compared to other devices in the market and submitted them to the FDA.
Data Science Lead
UMBA
- Led a team of data scientists and data engineers in charge of the lending decisions of the company.
- Took responsibility for the risk management models leveraging different machine learning techniques to predict probability of default and for the operational back-end Python API.
- Contributed to the portfolio of loans that grew on average 15% per month while keeping default constant in the data acquisition phase of the company.
Senior Data Scientist | CTO | Co-founder
SYNX.AI
- Created a tool for the automatic searching of the best combinations of preprocessing strategies, neural architectures, and hyperparameters using parallel training in several GPUs.
- Trained a computer vision model with deep convolutional neural networks for the diagnostic of diabetic retinopathy.
- Used machine learning to develop several financial risk models for different Mexican fintechs and banks, helping them to reduce the load on human analysts, adjust their risk strategies, and optimize their portfolios.
- Developed a machine-learning churn model for a major Mexican payment processing company.
- Developed a tool to monitor and optimize aircraft fuel spending for a major international airline based in Abu Dhabi.
- Created a tool for online advertising budget optimization based on reinforcement learning (contextual bandits).
- Performed as a sales engineer—talking with potential clients in order to understand their business needs and translate them into technical data science specifications.
- Lead a team of six data scientists—mentoring and supervising their progress in different projects, making sure everyone was engaged, learning and delivering the right results for our clients.
- Created a demand forecasting model for a large food company using the Prophet library.
Senior Data Scientist
Wizeline
- Designed and oversaw the creation of a tool for the understanding and prediction of oil price differentials arising from the interaction between production volumes, refinery demand and transportation costs, using geographic and financial data.
- Performed as the main technical contact with the client, an oil trading company based in Colorado.
- Led a team of four data scientists and one engineer.
Data Scientist in Computer Vision
Makeup on Us
- Developed and optimized face and facial landmark detectors, as well as a color synthesizer using transformations in different color spaces as key components of a makeup augmented reality system.
- Presented our technology in a talk at the Microsoft Reactor Center in San Francisco.
- Developed prototypes for other facial computer vision systems such as emotion recognition and face identification using deep convolutional neural networks.
Data Scientist in Financial Analysis
Kueski
- Created a graph database using Neo4J codifying several relationships among customers (phone, Facebook friends, addresses, and more) in order to create network features to feed a fraud detection machine learning model.
- Designed and trained a machine learning model to detect the probability of fraud (identity theft), which included features from a Neo4J graph database, that allowed a 50% reduction in the volume of applications human analysts had to review.
- Participated in the financial analysis of the company's portfolio, creating metrics and insights into the evolution of cohorts, profitability, and so on.
Data Scientist in Marketing
Linio
- Optimized eCommerce marketing by developing a model to monitor and calibrate TV advertising campaigns in Latin American using precise information about spot timings, costs and channels and measuring their impact on online visits.
- Used genetic algorithms to find the best possible configuration of agents in customer service call center, taking into consideration the historical hourly volume of calls and parameters such as desired occupancy and operational costs.
- Created a product recommender system based on visit and transaction data using Apache Spark.
Assistant Researcher in Neuroscience
University of Wisconsin
- Applied methods from complex dynamical systems theory such as synchronization and recurrent analysis to electroencephalographic data.
- Applied an independent component analysis for sensor data cleaning.
- Explored the consequences of using different information-theoretical methodologies such as mutual information in the understanding of chaotic systems.
Experience
Deep Learning for the Diagnosis of Diabetic Retinopathy
The model was trained to explore a vast space of convolutional neural network architectures inspired by Inception and ResNet (residual neural network) using best practices such as transfer learning, data augmentation, regularization, dropout, ensembles, and parallel training in several GPUs.
The system was designed to screen patients and take the workload off from ophthalmologists. The model had a sensitivity of 95% (true positive rate), while it had a specificity of 65% (true negative rate). This meant that the ophthalmologists only had to manually review 35% of the negative cases, resulting in much more efficient use of their time.
The system has not yet reached the stage of commercial distribution due to funding and regulatory issues.
Modeling Geographical Differences in the Price of Oil
Skills
Languages
Python, SQL, R, Cypher
Paradigms
Data Science
Other
Machine Learning, Data Mining, Data Queries, Financial Modeling, Mathematics, Artificial Intelligence (AI), Computer Vision, Natural Language Processing (NLP), Financial Data Analytics, Deep Learning, Unsupervised Learning, Credit Risk, Statistics, Facial Recognition, Convolutional Neural Networks (CNN), Object Detection, GPT, Generative Pre-trained Transformers (GPT), Data Analytics, Data Reporting, QGIS, Financial Markets, GeoPandas, Recommendation Systems, Risk Management, Data Engineering, Decentralized Finance (DeFi), Ethereum Smart Contracts, Web Scraping, OpenAI, OpenAI GPT-4 API, OpenAI GPT-3 API, LangChain, Neuroscience, Physics, Computer Science, Language Models, Prompt Engineering, ChatGPT, Retrieval-augmented Generation (RAG)
Libraries/APIs
Pandas, Scikit-learn, TensorFlow, Keras, NumPy, OpenGL, SciPy, NetworkX, XGBoost, OpenCV, Dlib, PySpark
Tools
Microsoft Excel, Spark SQL, Amazon SageMaker, MATLAB, Plotly, GIS, BigQuery, Mathematica
Storage
NoSQL, Neo4j, MongoDB, MySQL, Amazon S3 (AWS S3), Graph Databases, Amazon Aurora
Frameworks
Flask, Spark, LlamaIndex
Platforms
Amazon Web Services (AWS), Docker
Education
Master's Degree in Physics
Institute of Physics, UNAM - Mexico City, Mexico
Participated in a Research Stay in Neuroscientific Data Analysis
University of Wisconsin - Madison, WI, USA
Bachelor's Degree in Physics
National University of Mexico - Mexico City, Mexico
Participated in the Santader Scholarship Exchange Program in Physics
University of Madrid - Madrid, Spain
Certifications
AWS Cloud Practitioner
Amazon Web Services (AWS)
Certified TensorFlow Developer
TensorFlow Certificate Program
CFA Level I
CFA Institute
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