Francisco Lucio S. Bustamante

A chemist working with Data Science and Python Programming

Abstract

Senior Data Scientist specialized in building end-to-end ML solutions, from Big Data processing to deployment. Currently leveraging Generative AI to optimize recommendations at LuizaLabs (via TeclaT). Experience includes delivering analytics for Uber, predictive modeling for Climatempo (both via Necto Systems), and teaching Data Science. Former academic researcher with a strong scientific background.

Professional Experience

2025 - Present: Senior Data Scientist at LuizaLabs (via TeclaT)

TeclaT (Contract) — Assigned to LuizaLabs, the technology arm of Magazine Luiza (Magalu), a leading Brazilian omnichannel retailer, as a Senior Data Scientist in the Recommendations team, improving search and product discovery at scale.

  • Improved “Buy It Again” repurchase recommendations with up to ~10% Recall@10 lift and up to ~20% coverage increase (category-dependent) by shifting to a curated, entity-based consumables scope.
  • Built a config-driven Airflow DAG orchestrating PySpark jobs over Parquet datasets in GCP Cloud Storage (GCS), enabling per-category processing via JSON allow/block lists.
  • Implemented a PySpark UDF for deterministic entity normalization (slugification) to prevent config-to-catalog mismatches (casing/accents) at scale.
  • Developed notebook-based pipeline replay and auditing on production snapshots to validate entity gaps and quantify share-of-category coverage before deployment.
  • Built GenAI intent/semantic feature extraction for ranking models using vLLM + LangChain + Llama.
  • Queried and prepared large-scale data using BigQuery and PySpark; tracked experiments with MLflow and packaged workloads with Docker; collaborated via GitLab.

2025: Data Scientist at NECTO Systems

Worked at Climatempo, a major weather intelligence company in Brazil, via NECTO Systems. Delivered a full-cycle data science solution:

  • Replaced experience-based heuristics with a data-driven outage risk scoring framework, grounding decisions in meteorological signals rather than prior-year occurrence patterns alone.
  • Integrated weather time series (rainfall, wind, temperature) with grid/asset and historical incident data; engineered spatiotemporal features to capture storm intensity and exposure.
  • Trained and validated predictive models (e.g., tree-based ML) with time-based backtesting; delivered interpretable drivers to support operational decision-making.
  • Operationalized outputs into monitoring/decision workflows (dashboards/alerts), standardizing risk assessment and reducing subjectivity across shifts.
  • Delivered reproducible code with full documentation, assisted in technical transfer, and supported initial deployment in internal pipelines

Click here for a more detailed description of the project.

Additionally, worked on a short-term assignment at Ball Corporation, a global leader in sustainable aluminum packaging, where I:

  • Designed an analytics solution to integrate packaging-material inventories, production capacity, and sales plans into a unified planning view
  • Developed Python (pandas) and SQL ETL pipelines plus advanced Excel automation to consolidate and standardize multiple data sources
  • Built Power BI dashboards to monitor stock coverage and capacity vs. demand, reducing manual reporting effort and supporting data-driven decisions

2023 - 2025: Python and Data Science Instructor at Hashtag Treinamentos

Hashtag Treinamentos is one of Brazil’s largest online education providers, having trained over 80,000 students in Python and Data Science.

  • Led the creation and review of comprehensive teaching materials on Python libraries (NumPy, Pandas, Matplotlib, Seaborn, Scikit-Learn, TensorFlow, PyTorch, Streamlit, Plotly/Dash), ensuring up-to-date and industry-aligned content.
  • Produced video tutorials on SQL, Git, virtual environments, and Python best practices.
  • Improved the Net Promoter Score (NPS) of Data Science courses by 30% through continuous feedback loops and content optimization.
  • Introduced real-world projects to the curriculum, covering fraud detection, customer churn prediction, marketing campaign optimization, price optimization, and computer vision.
    • Developed 15 hands-on projects simulating practical industry scenarios.
  • Provided ongoing support to students by answering questions, reviewing code, and offering additional resources to facilitate comprehension of complex topics.

Click here for a more detailed description of this entry, with a list of projects specially tailored for educational purposes and that have a hands-on approach towards real-world applications.

2022 - 2023: Data Analyst at Uber (through NECTO Systems)

Allocated to Uber via NECTO Systems — a company specializing in bespoke software solutions — focusing on accounting reconciliation and improving fiscal processes.

  • Utilized SQL, Python (Pandas), Hive, and Presto for large-scale data analysis and processing.
  • Cleaned and standardized data from multiple sources, ensuring data accuracy and integrity.
  • Reviewed and optimized existing data pipelines, improving overall efficiency and reducing error rates by 20%.
  • Validated reconciled data for compliance with international accounting standards.
  • Assisted Uber in adopting compatible accounting practices and meeting fiscal requirements.
  • Implemented automated data search and reconciliation pipelines using data engineering techniques, reducing processing time by 15%.

Click here for a more detailed description of the project.

In addition to my project responsibilities, I contributed to NECTO Systems’ blog, authoring articles on the application of analysis and data science in industrial and business environments, thereby sharing insights on leveraging data for informed decision-making and operational efficiency.

2020 - Present: Science communicator at Ciência Programada

Created personal projects aimed at combining science communication with programming.

  • Published over 100 articles and created video content explaining programming concepts, Python libraries (Pandas, Matplotlib, etc.), and data-driven approaches to scientific problems.
  • Covered coding best practices, Git, and testing methodologies.
  • Attracted thousands of monthly visitors, broadening understanding of science and programming.

To see the project website click here (Brazilian Portuguese)

2015 - 2022: Researcher and Professor at IFRJ - Instituto Federal de Educação, Ciência e Tecnologia do Rio de Janeiro (Brazil)

  • Launched a research group on educational software for chemistry, leveraging Python for data visualization (Matplotlib, Plotly), data analysis (Pandas, NumPy, SciPy), and version control (Git).
  • Supervised 3 undergraduate research students, coordinating projects using Trello.
  • Delivered Physical Chemistry lectures to over 500 students, integrating hands-on experience with Excel and Python to boost employability skills.

To see the projects I’ve developed related to this experience click here

2014 - 2015: Chemist - Laboratory manager at UFF - Universidade Federal Fluminense (Brazil)

  • Oversaw the Central Analytical Facility, scheduling equipment usage and training.
  • Increased equipment availability by 100% through improved resource management.
  • Automated laboratory usage reports using Python and Excel, significantly streamlining workflows.
  • Designed and launched training programs for laboratory equipment operations and data analysis, training over 100 participants.

To see the projects I’ve developed related to this experience click here

2014: Environmental analyst at INEA - Instituto Estadual do Ambiente do Rio de Janeiro (Brazil)

  • Automated environmental indicator reports in Excel, improving team productivity by 25%.
  • Contributed to implementing ISO/IEC 17025 standards at the Environmental Analysis Laboratory.

To see the projects I’ve developed related to this experience click here

2012 - 2014: Researcher and Professor at UFF - Universidade Federal Fluminense (Brazil)

  • Supervised two projects focused on creating low-cost educational resources.
  • Taught chemistry classes and implemented exercises integrating real-world chemical data analysis.

Education

Licenses & certifications

  • 2020
    • Course: Python Pro
      • Skills: Python; Django framework; Test Driven Development (TDD); Heroku; Docker.
    • Course: Data Science Pro
      • Skills: Python data science tools such as Pandas; Jupyter Notebooks; Google Colab; Matplotlib.
  • 2019
    • Course: Data Science by Data Bootcamp
      • Skills: Python data science tools such as Pandas; Jupyter Notebooks; Google Colab; Matplotlib.
  • 2018
    • Introduction to Computer Science and Programming Using Python by MITx at edX

Academic experience

To see the projects I’ve developed during my education click here.

Master’s Degree in Chemistry, Universidade Federal Fluminense - 2010-2012

  • Conducted extensive data analysis using Excel and Origin.
  • Published three research papers in international peer-reviewed journals.
  • Served on organizing committees for two scientific symposia.

Bachelor’s Degree in Chemistry, Universidade Federal Fluminense - 2006-2010

  • Published one scientific paper in a national peer-reviewed journal.
  • Winner of the Alumni Laureate Award due to high academic performance