Data Scientist

Rylee is an e-commerce platform designed to assist customers with product insights and market analysis to improve their market strategies on Bol.com and Amazon.

AI Tools: Achieved Databricks certification on Generative AI, giving professional recognition for working with RAG, and Agent models.

Machine Learning Time Series: Worked on a large-scale sales forecasting model harvesting Rylee’s own data plus scraped data from Bol.com. This model aims to predict sales of products on Rylee’s database as well as the best seller products on Bol.com.

API: Created an API to handle product queries, giving insights and sales forecasting using Flask and serverless frameworks AWS Lambda.

ETL: Designed and implemented ETL pipelines using dbt, Airflow, Spark to orchestrate and automate the feature engineering process, which included an async solution to retrieve data from Bol.com API efficiently, respecting the rate limits, but at the same time parallelizing the process at scale. Furthermore, this process included automating data reconciliation from multiple sellers across Rylee and Bol.com.

Optimization: Utilized Pytorch and Spark for advanced machine learning to optimize high-performance models.

Homemove is a comprehensive platform offering various moving-related services, including surveys, removals, and mortgages, all integrated within a single app.

AI tools: Contributed to developing an LLM-powered negotiation tool, which allows users to obtain quotes and negotiate prices through an AI chatbot, add customers to our CRM system automatically, and alert the sales team upon successful negotiation. Used Open AI Assistant and GPT models.

Data Transformation: Led and established a scalable data transformation initiative, leveraging Snowflake for cloud data warehousing and Sigma for BI and visualization.

ETL: Designed and implemented ETL pipelines using Snowflake, Python, SQL, dbt, and Airflow from scratch to automate data ingestion and transformation processes.

Machine Learning: Developed a predictive modelling solution that reduces marketing costs and improves targeting by identifying high-potential home movers.

Optimization: Utilized Pytorch and Snowpark for advanced machine learning to optimize highperformance models.

Achievements: This predictive model will be used to attract investment during Homemove's Series A funding in February.

FourthRev is a company that specializes in creating education-to-employment pathways in collaboration with leading universities and tech companies. They focus on delivering career-relevant education and their programs combine the academic rigor of university with practical, real-world skills and knowledge. He worked as a Data Science specialist creating and teaching data science to Cambridge University students.

Felipe was recruiter by Cambridge University to develop and teach an advanced online data science course at Cambridge University, which included a vast implementation and application of machine learning theories from scratch.

During this project, he had demonstrated his deep technical understanding of data science and machine learning as well as the ability to maintain the highest standard of communication of technical knowledge.

He used his experience and expertise to create a high-standard curriculum with complexity and depth, including fundamental topics such as Neural Networks, NLP for AI, Unsupervised Learning, and several nuances of Decision Trees, from basic ideas to the complex algorithms of XGBoost.

Employed innovative teaching methods to enhance student engagement and learning outcomes.

Recognized by academic peers for my understanding of machine learning topics and effective teaching methodologies in data science education.

Outra is a data-driven property insight company that specialises in providing clients with targeted data at a household level to focus their resources and delivery of their services.

Felipe worked on the migration from Dataiku to a new custom Intelligence fabric developed in-house using tools such as MLflow for tracking, Airflow for orchestration, Snowflake as a database, GitHub Actions for CI/CD, AWS for storing and clustering, and DBT for transformations (Data Engineering).

Created two major models used in the company that predict whether a household will list for sale/rent and when it will be taken off the market (sold/rented), forecasting the full household sale life cycle. These models single-handedly generated a major multi-million pound partnership with Zoopla.

Use of LLMs as assistants for code documentation, coding co-pilot, provide customers with an interactive chatbot for our dashboards and data shown to them.

He created his own ETL/ELT pipelines to transform raw data and prepare it for modelling.

Utilized advanced querying, visualization, and analytics tools to such as KeplerGI, Seaborn, and Dataiku to create maps and diagrams for non-technical users to visualise a certain problem.

Belmont Green is a specialist mortgage lending company turned bank, focusing on providing financial and mortgage solutions to financially affected customers.

Responsible for creating a conversion model using survival analysis techniques, and responsible for taking the project from beginning to production.

Building Machine Learning algorithms and statistical models for time series data, focusing on retention, lifetime value, and expected loss models.

Taking ownership of projects from start to finish to ensure proofs of concept were properly implemented and deployed in production.

Implement Machine Learning algorithms for a variety of tasks such as Cashflow models, Early redemption models, Default models, Pre-payment models, and Conversion models, using Python and R.

Perform clustering and segmentation algorithms to gain insights into the usage and appeal of various products and features for marketing and other purposes.

Boster.Ai is a company dedicated to create No-code bots for data retrieval, monitoring and automation. Originally, they started as an IT consultancy company, creating personalized solutions to small and mid-size companies to harvest the power of Machine Learning

Felipe worked performing data exploration, analysis, and building Machine Learning algorithms and statistical models for several start-ups/mid-size companies in the UK and USA.

He built and diagnostic Neural Networks models for forecasting key performance indicators using Python, TensorFlow, and Keras.

Worked in e-commerce businesses solving customer behaviour problems such as lifetime value, clustering of customers, etc.

Performed ethical web scraping using Python with scraPy, RoboBrowser, and BeautifulSoup to obtain data for various analyses.