Data science has become a profession with expectations to match. Analysts who once worked alone in notebooks now contribute to shared codebases, defend findings in front of risk committees and auditors, and hand models to engineering teams who must run them for years. Weak statistical reasoning, untested code, unexplained models and overconfident forecasts are no longer tolerated, particularly in regulated sectors such as banking, insurance, health and public administration.

The Professional Data Science & Machine Learning in Practice Training Course is written for practitioners who can already build a model and want to do so to a higher standard. The first five days strengthen the foundations of professional work: scoping engagements, writing maintainable and tested Python, handling data at scale, statistical inference and a disciplined modelling toolkit. The second five days go further into calibration and uncertainty quantification, causal inference and experimentation, explainability, fairness and professional ethics, and production delivery, finishing with a capstone that is code-reviewed and presented as a client engagement.

Participants work in a shared Git repository from the first morning, review one another's code and write short technical memos as well as notebooks. Exercises draw on financial services, health, energy and public sector data, and the emphasis throughout is on defensible choices: why a method was chosen, what its assumptions are and how confident anyone should be in the answer.

After the ten days, experienced practitioners will be able to:

• Scope a data science engagement with clear deliverables, assumptions, risks and acceptance criteria.
• Write modular, tested and version-controlled Python that others can review and maintain.
• Process large datasets efficiently with SQL, DuckDB, Polars and PySpark.
• Apply rigorous inference, including bootstrapping, power analysis and Bayesian estimation.
• Quantify predictive uncertainty through calibration, prediction intervals and conformal methods.
• Estimate causal effects from observational and experimental data and state the assumptions behind them.
• Document, explain and audit models for fairness using model cards, SHAP and Fairlearn.
• Deliver models through reproducible pipelines with automated testing, deployment and monitoring.

Participants should already build models in Python. Typical roles include:

• Data scientists with one or more years of hands-on experience
• Senior data analysts and analytics engineers moving into modelling
• Statisticians and econometricians in central banks, statistics bureaus and research institutes
• Quantitative analysts and model developers in banking and insurance
• Machine learning engineers seeking stronger statistical grounding
• Model validators and internal audit specialists reviewing analytical work
• Health and epidemiological data scientists
• Data science consultants delivering client projects
• Lead analysts in development agencies running impact evaluation and targeting models

Participants return to work with:

• A project template repository with environment files, tests, linting and continuous integration configured.
• An engagement scoping document and stakeholder plan they can reuse with internal or external clients.
• A statistical analysis plan covering power, multiple comparisons and uncertainty reporting.
• A calibrated model with conformal prediction intervals and a decision-threshold analysis.
• A causal effect estimate with a documented causal diagram and robustness checks.
• A completed model card and fairness audit for a model of their own.
• Written feedback from a structured code review of their capstone submission.
• A personal development plan listing target competencies and the evidence to collect for each.

The programme treats participants as working professionals and mirrors how strong data teams operate. It uses:

• A shared Git repository in which all lab work is committed, reviewed and tested
• Expert briefings that focus on assumptions, failure modes and judgement calls
• Advanced labs on financial, health, energy and public sector datasets
• Structured peer code reviews and critique of statistical write-ups
• Written technical memos alongside notebooks to build reporting discipline
• A capstone run as a simulated client engagement with a review panel

Participants who attend every session, commit their lab work to the course repository and present a reviewed capstone receive a CPD-accredited Certificate of Completion issued by Vision Reach Global Consultancy.

The certificate states the course title, dates and mode of delivery. The capstone repository, model card and technical memo together form a professional portfolio that participants can show to employers or clients.