Data Scientist
Accountable for framing a business decision as a modelling problem and for building, validating and explaining predictive or causal models fit for that decision.
1task in this job
22skills it uses
5tools named
The skills this job leans on most
- 1Training reproducible machine learning modelsNeeded in 1 of 1 tasks
- 2Evaluating ML generalisation and fitnessNeeded in 1 of 1 tasks
- 3Choosing an analytical method to fit the question and dataNeeded in 1 of 1 tasks
ToolsMLflowJupyter (Notebook / JupyterLab)OptimizelyAmplitudePythonpandasscikit-learnDatabricksR (with RStudio / Posit)Hugging Face
All 1 tasks, where people get stuck and all 22 skills
Tasks
Where people get stuck
- Evaluating experimental results and integrity
- Estimating and challenging causal effects
- Weighing conflicting evidence by design strength
All skills
- Analysing event sequences, states and durations
- Assessing data source credibility and fitness for a question
- Calculating and interpreting percentages, rates and ratios
- Choosing and justifying missing-data treatments
- Comparing groups
- Decomposing one's own tasks for AI assistance
- Describing distributions with summaries and plots
- Designing and interpreting hypothesis tests
- Designing efficient and reproducible investigations
- Diagnosing incorrect query and calculation results
- Eliciting the decision behind an analytical request
- Estimating quantities and differences with intervals
- Exploring data to find patterns that survive validation
- Gating releases on evidence with rollback ready
- Investigating anomalies
- Judging whether a change exceeds normal variation
- Profiling data
- Reading and tracing unfamiliar code
- Reading charts and dashboards correctly, including their state
- Selecting charts for evidence and decisions
- Standardising and cleaning data under agreed rules
- Writing evidence-led findings and auditable recommendations with their limits
Paths to start
Skill Path
Exploratory Data Analysis and Pattern Discovery
Describe a distribution honestly and decide which unusual records need checking before anyone acts.
- Modules
- 5 open
- Practice data
- Raket Parts
Tool Path
SQL for Analysis
Answer real business questions with SQL, from your first query to finding why a number moved.
- Modules
- 5 open
- Runs in
- Your browser
Coming soonMachine Learning: from Data to Validated Models and Decisions
6tasks in this job
90skills it uses
10tools named
The skills this job leans on most
- 1Training reproducible machine learning modelsNeeded in 6 of 6 tasks
- 2Evaluating ML generalisation and fitnessNeeded in 4 of 6 tasks
- 3Choosing an analytical method to fit the question and dataNeeded in 4 of 6 tasks
- 4Evaluating experimental results and integrityNeeded in 1 of 6 tasks
- 5Estimating and challenging causal effectsNeeded in 1 of 6 tasks
ToolsMLflowJupyter (Notebook / JupyterLab)OptimizelyAmplitudePythonpandas+ 4 more
All 6 tasks, where people get stuck and all 90 skills
Tasks
- Monitor a deployed model or AI system in use and recommend retraining, recalibration, change or retirement
- Prepare a model or AI system release decision with evaluation evidence, monitoring and rollback
- Design an ethical decision-relevant experiment
- Develop and document a fit-for-purpose analytical or ML model
- Evaluate statistical and causal evidence for an intervention's impact
- Develop and evaluate an LLM-based application against an agreed evaluation bar
Where people get stuck
- Evaluating experimental results and integrity
- Evaluating ML generalisation and fitness
- Choosing an analytical method to fit the question and data
All skills
- Choosing an analytical method to fit the question and data
- Estimating and challenging causal effects
- Evaluating ML generalisation and fitness
- Evaluating experimental results and integrity
- Training reproducible machine learning models
- Analysing error patterns and failure modes in models and AI systems
- Analysing event sequences, states and durations
- Applying research ethics and participant consent
- Assessing data source credibility and fitness for a question
- Assessing data-driven fairness and bias
- Assessing privacy risk through data protection impact assessments
- Assessing supplier dependency, data-handling and exit risk
- Bounding and securing AI systems' tool and data access
- Building data protection by design into data products and pipelines
- Building decision-focused reports and dashboards
- Calculating and interpreting percentages, rates and ratios
- Choosing and justifying missing-data treatments
- Classifying AI uses against regulatory obligations
- Combining tables safely: cardinality, fan-out and conservation of totals
- Communicating quantified uncertainty
- Communicating sensitive or unwelcome analytical findings
- Comparing groups
- Constructing clear and honest charts
- Controlling disclosure risk in shared data and outputs
- Decomposing one's own tasks for AI assistance
- Defining measurable outcomes and leading indicators
- Defining the analysis population: filters, nulls, dates and time zones
- Describing distributions with summaries and plots
- Designing AI-assisted workflows with bounded tools and human authority
- Designing and appraising samples
- Designing and evaluating retrieval for AI systems
- Designing and interpreting hypothesis tests
- Designing and quality-checking survey instruments
- Designing and running evaluations of AI systems
- Designing decision-oriented research with customers, users or employees
- Designing efficient and reproducible investigations
- Designing experiments
- Designing implementable change plans
- Diagnosing capability gaps from work evidence and planning development
- Diagnosing incorrect query and calculation results
- Diagnosing misleading data visualisations
- Disclosing AI contribution to professional work
- Eliciting the decision behind an analytical request
- Establishing and declaring data grain
- Estimating quantities and differences with intervals
- Evaluating and improving chart accessibility
- Evaluating employee experience evidence responsibly
- Evaluating fitness and full cost of AI-assisted work
- Evaluating learning transfer and performance outcomes
- Evaluating responsible AI use risks
- Evaluating whether people understood, adopted and sustained a change
- Explaining technical work to accountable non-specialists
- Exploring data to find patterns that survive validation
- Fitting, checking and interpreting inferential regression models
- Formulating answerable data questions
- Gating releases on evidence with rollback ready
- Interpreting averages, weighted measures and mix effects
- Interpreting correlation without inferring cause
- Interpreting customer feedback and qualitative evidence
- Interpreting trends and seasonality
- Investigating anomalies
- Judging the materiality of defects and deficiencies for the intended use
- Judging whether a change exceeds normal variation
- Judging whether and how to use AI for a task
- Judging whether uncertainty changes the decision
- Measuring campaign outcomes with valid denominators
- Monitoring model fitness and drift
- Preparing analysis-ready datasets reproducibly
- Presenting and defending evidence-based recommendations to decision-makers
- Profiling data
- Protecting sensitive data in AI-assisted work
- Rating risks against appetite and choosing proportionate treatment
- Reading and tracing unfamiliar code
- Reading charts and dashboards correctly, including their state
- Recognising and resolving ethical conflicts in data and AI work
- Recognising misleading numerical claims
- Selecting charts for evidence and decisions
- Setting and applying a release and rollback bar for models and AI systems
- Setting and reviewing action thresholds for scores and model predictions
- Specifying and iterating AI task instructions
- Standardising and cleaning data under agreed rules
- Statistical power and sample size
- Testing competing explanations with discriminating evidence
- Testing, implementing and sustaining measured improvements
- Understanding and interrogating data and application APIs
- Verifying AI-assisted outputs against sources
- Verifying benefits realisation and attribution
- Weighing conflicting evidence by design strength
- Writing evidence-led findings and auditable recommendations with their limits
- Writing maintainable code for analytical and data work
Paths to start
Skill Path
Exploratory Data Analysis and Pattern Discovery
Describe a distribution honestly and decide which unusual records need checking before anyone acts.
- Modules
- 5 open
- Practice data
- Raket Parts
Tool Path
SQL for Analysis
Answer real business questions with SQL, from your first query to finding why a number moved.
- Modules
- 5 open
- Runs in
- Your browser
Skill Path
Experimentation and Causal Impact Evaluation
Answer "did it work" with randomised experiments where possible and quasi-experimental methods where not.
- Modules
- 5 open
- Practice data
- Raket Parts
Coming soonMachine Learning: from Data to Validated Models and Decisions
7tasks in this job
97skills it uses
10tools named
The skills this job leans on most
- 1Training reproducible machine learning modelsNeeded in 6 of 7 tasks
- 2Evaluating ML generalisation and fitnessNeeded in 4 of 7 tasks
- 3Choosing an analytical method to fit the question and dataNeeded in 4 of 7 tasks
- 4Evaluating experimental results and integrityNeeded in 1 of 7 tasks
- 5Estimating and challenging causal effectsNeeded in 1 of 7 tasks
ToolsMLflowMicrosoft Power BIWeights & BiasesOptimizelyPythonscikit-learn+ 4 more
All 7 tasks, where people get stuck and all 97 skills
Tasks
- Monitor a deployed model or AI system in use and recommend retraining, recalibration, change or retirement
- Prepare a model or AI system release decision with evaluation evidence, monitoring and rollback
- Design an ethical decision-relevant experiment
- Develop and document a fit-for-purpose analytical or ML model
- Evaluate statistical and causal evidence for an intervention's impact
- Develop and evaluate an LLM-based application against an agreed evaluation bar
- Set and maintain working standards for a team's analytical, reporting and modelling work products
Where people get stuck
- Evaluating experimental results and integrity
- Evaluating ML generalisation and fitness
- Choosing an analytical method to fit the question and data
All skills
- Choosing an analytical method to fit the question and data
- Estimating and challenging causal effects
- Evaluating ML generalisation and fitness
- Evaluating experimental results and integrity
- Training reproducible machine learning models
- Analysing error patterns and failure modes in models and AI systems
- Analysing event sequences, states and durations
- Applying data classification and handling rules
- Applying research ethics and participant consent
- Assessing data source credibility and fitness for a question
- Assessing data-driven fairness and bias
- Assessing privacy risk through data protection impact assessments
- Assessing supplier dependency, data-handling and exit risk
- Bounding and securing AI systems' tool and data access
- Building data protection by design into data products and pipelines
- Building decision-focused reports and dashboards
- Calculating and interpreting percentages, rates and ratios
- Choosing and justifying missing-data treatments
- Classifying AI uses against regulatory obligations
- Combining tables safely: cardinality, fan-out and conservation of totals
- Communicating quantified uncertainty
- Communicating sensitive or unwelcome analytical findings
- Comparing groups
- Constructing clear and honest charts
- Controlling disclosure risk in shared data and outputs
- Decomposing one's own tasks for AI assistance
- Defining measurable outcomes and leading indicators
- Defining the analysis population: filters, nulls, dates and time zones
- Describing distributions with summaries and plots
- Designing AI-assisted workflows with bounded tools and human authority
- Designing and appraising samples
- Designing and evaluating retrieval for AI systems
- Designing and interpreting hypothesis tests
- Designing and quality-checking survey instruments
- Designing and running evaluations of AI systems
- Designing decision-oriented research with customers, users or employees
- Designing efficient and reproducible investigations
- Designing experiments
- Designing implementable change plans
- Diagnosing capability gaps from work evidence and planning development
- Diagnosing incorrect query and calculation results
- Diagnosing misleading data visualisations
- Disclosing AI contribution to professional work
- Eliciting the decision behind an analytical request
- Escalating unresolved risks with evidence and clear ownership
- Establishing and declaring data grain
- Estimating quantities and differences with intervals
- Evaluating and improving chart accessibility
- Evaluating employee experience evidence responsibly
- Evaluating fitness and full cost of AI-assisted work
- Evaluating learning transfer and performance outcomes
- Evaluating responsible AI use risks
- Evaluating whether people understood, adopted and sustained a change
- Explaining technical work to accountable non-specialists
- Exploring data to find patterns that survive validation
- Fitting, checking and interpreting inferential regression models
- Formulating answerable data questions
- Gating releases on evidence with rollback ready
- Giving specific developmental feedback on data and AI work
- Independently checking analytical conclusions and claims
- Interpreting averages, weighted measures and mix effects
- Interpreting correlation without inferring cause
- Interpreting customer feedback and qualitative evidence
- Interpreting trends and seasonality
- Investigating anomalies
- Judging the materiality of defects and deficiencies for the intended use
- Judging whether a change exceeds normal variation
- Judging whether and how to use AI for a task
- Judging whether uncertainty changes the decision
- Measuring campaign outcomes with valid denominators
- Monitoring model fitness and drift
- Preparing analysis-ready datasets reproducibly
- Presenting and defending evidence-based recommendations to decision-makers
- Profiling data
- Protecting sensitive data in AI-assisted work
- Rating risks against appetite and choosing proportionate treatment
- Reading and tracing unfamiliar code
- Reading charts and dashboards correctly, including their state
- Recognising and resolving ethical conflicts in data and AI work
- Recognising misleading numerical claims
- Reviewing human- and AI-generated code changes for correctness and risk
- Sampling and reviewing a team's work against standards
- Selecting charts for evidence and decisions
- Setting and applying a release and rollback bar for models and AI systems
- Setting and reviewing action thresholds for scores and model predictions
- Specifying and iterating AI task instructions
- Standardising and cleaning data under agreed rules
- Statistical power and sample size
- Testing competing explanations with discriminating evidence
- Testing, implementing and sustaining measured improvements
- Understanding and interrogating data and application APIs
- Verifying AI-assisted outputs against sources
- Verifying benefits realisation and attribution
- Verifying reports and dashboards before and after release
- Weighing conflicting evidence by design strength
- Writing evidence-led findings and auditable recommendations with their limits
- Writing maintainable code for analytical and data work
All tools
- MLflow
- Microsoft Power BI
- Weights & Biases
- Optimizely
- Python
- scikit-learn
- Jupyter (Notebook / JupyterLab)
- R (with RStudio / Posit)
- Hugging Face
- Git / GitHub
Paths to start
Skill Path
Exploratory Data Analysis and Pattern Discovery
Describe a distribution honestly and decide which unusual records need checking before anyone acts.
- Modules
- 5 open
- Practice data
- Raket Parts
Tool Path
SQL for Analysis
Answer real business questions with SQL, from your first query to finding why a number moved.
- Modules
- 5 open
- Runs in
- Your browser
Skill Path
Experimentation and Causal Impact Evaluation
Answer "did it work" with randomised experiments where possible and quasi-experimental methods where not.
- Modules
- 5 open
- Practice data
- Raket Parts
Coming soonMachine Learning: from Data to Validated Models and Decisions
About the role
Accountable for framing a business decision as a modelling problem and for building, validating and explaining predictive or causal models fit for that decision. Chooses methods, validation design and the evidence needed to trust a model. Differs from a Machine Learning Engineer, who owns running models in production.
Also called Statistician