Evidence before answers
Learners inspect source material and document uncertainty before proposing features.
CASE STUDY PRODUCT ROADMAP
Build the mission library as the core product.
A practical plan for turning seven case-study concepts into realistic, assessable deployment simulations with evidence rooms, working builds, staged complications, and portfolio outcomes.
01 / PRODUCT OBJECTIVE
Teach judgment through deployment evidence.
The mission library should be the central learning product—not a collection of decorative case descriptions. Each mission must recreate the conditions under which forward deployed engineers actually work: incomplete information, contradictory accounts, technical constraints, organizational resistance, and a decision window that keeps moving.
A completed mission should answer one question: Can the learner enter an unfamiliar operating environment, understand the real problem, design a credible intervention, deploy it responsibly, and prove whether it created value?
Decisions before dashboards
The technical build must improve a named operating decision, not merely visualize data.
Complications after commitment
New facts arrive after the learner has committed to an initial frame and design.
Portfolio proof
Every mission ends in inspectable artifacts, a working system, and an executive recommendation.
02 / BUILD SEQUENCE
Three missions establish the standard.
Build Atlas, Northstar, and Titan deeply before expanding the remaining four. Together they cover general FDE delivery, industrial analytics, and governed AI deployment.
01
FLAGSHIP LIVE
Atlas Freight
End-to-end operational deployment across logistics data, financial exposure, adoption, and value measurement.
PRIMARY CAPABILITYGeneral FDE judgment
Open mission →02
IN PRODUCTION
Northstar Manufacturing
Industrial analytics, data provenance, traceability, and recommendations under uncertain causal evidence.
PRIMARY CAPABILITYEvidence quality and analysis
View production specification →03
IN PRODUCTION
Titan Industrial
Enterprise RAG, document governance, safety controls, source hierarchy, and human escalation.
PRIMARY CAPABILITYProduction AI deployment
View production specification →04
PLANNED
Meridian Retail
Forecasting, optimization, financial tradeoffs, and local-manager overrides.
PRIMARY CAPABILITYRecommendations and adoption
View preview →05
PLANNED
GridWest Energy
Real-time prioritization, geospatial operations, resilience, and critical-infrastructure fairness.
PRIMARY CAPABILITYHigh-pressure operations
View preview →06
PLANNED
Apex Insurance
Responsible AI, auditability, urgent-case detection, and bias investigation.
PRIMARY CAPABILITYGoverned decision support
View preview →07
PLANNED
Harbor Health
Privacy-aware capacity coordination, clinical constraints, and allocation fairness.
PRIMARY CAPABILITYConstrained implementation
View preview →03 / STANDARD ARCHITECTURE
Every mission follows eight stages.
A consistent structure makes missions easier to build, easier for learners to navigate, and easier for reviewers to score without making the underlying problems predictable.
Orientation
Company profile, operating context, initial request, stated objective, timeline, and known constraints.
Initial request is intentionally incomplete.Evidence room
Emails, interviews, schemas, data extracts, policies, diagrams, incidents, and financial assumptions.
Evidence conflicts in meaningful ways.Discovery checkpoint
Stakeholder map, workflow, decision map, problem statement, constraints, and open questions.
Learner commits before full disclosure.Solution design
Solution thesis, requirements, architecture, data model, controls, risks, and measurement plan.
Scope reduction is rewarded.Build
Dashboard, API, pipeline, recommendation engine, RAG workflow, or other operational system.
Failures and security are part of the build.Hidden complication
A material change invalidates part of the original plan and forces a documented response.
The complication changes the design.Deploy and adopt
Pilot charter, release plan, rollback, runbook, training, support, ownership, and adoption metrics.
The system must survive beyond the demo.Value review
Mixed pilot evidence requires a scale, revise, narrow, continue, or stop recommendation.
Usage alone is not treated as value.04 / PRODUCTION ROADMAP
A twelve-week path to three complete missions.
The schedule prioritizes reusable infrastructure first, then one flagship mission, one industrial-data mission, and one AI mission.
PHASE 1
Create the mission system
Weeks 1–2
- Standard page and file structure
- Evidence-room and stakeholder components
- Checkpoint and complication patterns
- Shared rubric and reviewer guide
- Versioning and progress model
EXIT CRITERIAOne reusable structure that can support every future mission.
PHASE 2
Complete Atlas Freight
Weeks 3–6
- Customer environment and six stakeholders
- Messy CSV, JSON, email, and workbook evidence
- React, Python, and DuckDB starter build
- Identity-feed complication
- Pilot results and scoring examples
EXIT CRITERIAA fully runnable flagship mission ready for pilot learners.
PHASE 3
Build Northstar Manufacturing
Weeks 7–9
- Production-run data model
- Sensor, defect, maintenance, and material evidence
- On-premises architecture constraints
- Calibration-history complication
- Uncertainty-focused rubric
EXIT CRITERIAA mission where evidence quality materially affects the recommendation.
PHASE 4
Build Titan Industrial
Weeks 10–12
- Document corpus and authority hierarchy
- RAG starter and evaluation harness
- Safety and abstention controls
- Unsafe-notes complication
- AI governance scoring guide
EXIT CRITERIAA production AI mission that sometimes must refuse to answer.
01Framework
02Assessment
03Atlas context
04Atlas evidence
05Atlas build
06Atlas pilot
07Northstar context
08Northstar build
09Northstar assessment
10Titan context
11Titan build
12Titan pilot
05 / MISSION PACKAGES
One case, three levels of support.
The same operating environment can serve beginners, experienced engineers, certification candidates, and company teams by changing the amount of guidance and review.
Guided Mission
For learners building their first deployment case.
- Full evidence room
- Structured checkpoints
- Artifact templates
- Hints and concept explanations
- Reference debrief after completion
Independent Mission
For experienced engineers and interview preparation.
- Customer brief and evidence
- Data and constraints
- Submission requirements
- Published assessment rubric
- No solution guidance
Assessed Mission
For portfolio review and credential eligibility.
- Independent mission access
- Human portfolio review
- Written scoring and feedback
- Live technical defense
- One revision opportunity
06 / CONTENT ACCESS
Publish the promise. Protect the assessment.
Public pages should prove quality without exposing the full answer path or compromising assessed submissions.
Show openly
- Mission overview and learning outcomes
- Stakeholder summaries
- Sample evidence
- Required deliverables
- Assessment dimensions
- Portfolio outcomes
Gate after enrollment
- Full datasets and evidence room
- Complete source documents
- Hidden complication
- Pilot-result package
- Starter repository
- Structured checkpoints
Restrict tightly
- Ideal discovery findings
- Known traps and scoring anchors
- Architecture tradeoff notes
- Automatic-failure conditions
- Oral-defense questions
- Reference implementation
07 / QUALITY GATE
Do not publish a mission until it passes.
The standard prevents the library from becoming a set of clean tutorials with obvious answers.
The initial customer request is not the final problem statement.
Stakeholders disagree in ways that affect the design.
Data contains realistic defects and provenance issues.
The learner must make consequential tradeoffs.
The hidden complication changes the solution materially.
The technical build improves a named operating decision.
Deployment, adoption, and ownership affect success.
Technical, operational, and business value can be measured.
More than one defensible solution is possible.
The rubric separates weak, acceptable, and excellent work.
A reviewer can verify authorship in a live defense.
The finished work is credible portfolio evidence.
NEXT BUILD DECISION
Complete Atlas. Pilot it. Then expand.
The reusable system should be validated with real learners before all seven missions are built.