Project Roadmap¶
This roadmap translates the 10-week G.U.A.R.D. delivery plan into five dated milestones, each with defined sub-tasks and workload expectations.
Milestone Plan¶
| Phase | Activity and Sub-tasks | Dates | Workload |
|---|---|---|---|
| #1 Requirement Formalization | Investigation of Legal-to-Technical Mapping
|
April 20 to May 1 | 65-75 hours |
| #2 IR Design and Schema | Development of the Source of Truth
|
May 4 to May 15 | 70-80 hours |
| #3 Compliance Compiler | Logic Engine Development
|
May 18 to June 5 | 110-130 hours |
| #4 Integration and Enforcement | Shift-Left Implementation
|
June 8 to June 12 | 35-40 hours |
| #5 Evaluation and Synthesis | Validation and Final Documentation
|
June 15 to June 26 | 75-85 hours |
Notes¶
- Phase sequencing intentionally leaves weekday transition gaps for milestone closeout and planning handoff.
- Detailed execution status is maintained through linked GitLab milestone work items.
Research-Question Mapping¶
The roadmap phases contribute to one parent research question through supporting sub-questions.
| Milestone | Primary Question Contribution | Deliverable Type |
|---|---|---|
| #1 Requirement Formalization | SQ0 Existing Work, SQ1 Meta-Model | Problem framing, baseline comparison, IR attribute requirements |
| #2 IR Design and Schema | SQ1 Meta-Model, SQ2 Compiler Mapping | IR schema structure and mapping design constraints |
| #3 Compliance Compiler | SQ2 Compiler Mapping | Prototype translation and validation evidence |
| #4 Integration and Enforcement | SQ2 Compiler Mapping, SQ3 Impact | Pipeline integration and feedback-loop evidence |
| #5 Evaluation and Synthesis | Main Question synthesis from SQ0-SQ3 | Final argumentation, limitations, and recommendations |
Key Dependency Risk¶
If a high-impact clause cannot be made reliably machine-checkable, automation coverage can drop.
Mitigation strategy:
- Escalate interpretation to compliance/audit owners.
- Introduce compensating controls (manual gate, approval step, sampled audit evidence).
- Tag clause as human-in-the-loop and revisit formalization in subsequent iterations.
This risk is most sensitive to outputs from milestone #1 and milestone #2.