Data science analytics
Data Science Analytics
tasks: 21 constraints: 7 team: 6 timesteps: 50
Workflow Goal
Objective
Objective: Deliver a production‑ready analytics/modeling solution for a business problem with high data quality, responsible‑AI controls, and reproducible results.
Primary deliverables
- Curated and governed dataset with documented lineage and quality checks Feature pipeline and model artifacts with experiment tracking and seeds Model card, bias/fairness analysis, and explainability report Deployment package (CI/CD) with monitoring and rollback plan Executive readout with business impact, risks, and next steps
Acceptance criteria (high‑level)
- Data privacy/PII policy compliance; secrets not present in artifacts Reproducible training runs (fixed seeds, environment captured)
- Minimum evaluation thresholds met (e.g., AUC/accuracy and calibration)
- Bias/fairness metrics within policy thresholds and mitigations documented Deployment readiness gate passed (security review + monitoring plan)
Team Structure
| Agent ID | Type | Name / Role | Capabilities |
|---|---|---|---|
| data_engineer | ai | Ingestion pipelines Data quality checks Lineage documentation Data security |
|
| data_scientist | ai | Feature engineering Baseline modeling Experimentation |
|
| mlops_engineer | ai | Reproducibility CI/CD Packaging Monitoring |
|
| analytics_analyst | ai | KPI/ROI analysis Executive readouts |
|
| security_reviewer | human_mock | Security Reviewer (Security Review) | Dependency/secrets scans Deployment gate |
| risk_officer | human_mock | Risk Officer (Responsible AI Governance) | Fairness Explainability Governance |
Join/Leave Schedule
| Timestep | Agents / Notes |
|---|---|
| 0 | data_engineer — Ingestion & data quality data_scientist — Feature engineering & baseline |
| 10 | mlops_engineer — Reproducibility & experiment tracking |
| 20 | analytics_analyst — KPI analysis & readout drafting |
| 40 | risk_officer — RAI review: fairness & explainability |
| 50 | security_reviewer — Security review & deployment gate |
Workflow Diagram
Preferences & Rubrics
Defined: Yes.
Sources
- Workflow:
/Users/charliemasters/Desktop/deepflow/manager_agent_gym/examples/end_to_end_examples/data_science_analytics/workflow.py - Team:
/Users/charliemasters/Desktop/deepflow/manager_agent_gym/examples/end_to_end_examples/data_science_analytics/team.py - Preferences:
/Users/charliemasters/Desktop/deepflow/manager_agent_gym/examples/end_to_end_examples/data_science_analytics/preferences.py