The purpose of this narrative is to explain the ReviewShield Critical Path Method (CPM) and how it can be applied to the ReviewShield project using an Activity on Node (AON) diagram.
The ReviewShield Phase 1 AON is presented below. The tasks represent key actions and dependencies that need to be completed. Please note, the duration estimates below are based on the best-guess efforts and not the actual work efforts
Tasks and Dependencies
A: Define Requirements
A.1: Develop Customer-centric and Business Outcome focused SMART Goal
Duration: 2 weeks
Dependencies: None
A.2: Develop Success Metrics and reporting KPIs
Duration: 2 weeks
Dependencies: A.1
A.3: Develop a KPI reporting Spreadsheet Prototype and high-level resource estimation
Duration: 2 Weeks
Dependency: A.1
A.4: Develop high-level Product Requirement Narrative Stakeholder review, and product backlog grooming / Epic distribution strategy / Sprint plan for 12 weeks
Duration: 4 Weeks
Dependency: A.2
A.5: Develop Manual Audit SOP, Audit Cadence, and data analysis strategy
Duration : 3 weeks
Dependency: A.2
A.6: Establish a Progress Reporting mechanism such as routine email updates via an email distribution list, business documents, and stakeholder presentations based on pre-approved / standard templates
Duration : 3 weeks
Dependency: A.2
A.7: Finalize resource allocation and capacity management strategy to deliver P0 / MVP / V1 product and sprint cadence aligned with delivery schedule
Duration : 3 weeks
Dependency: A.2
B: Develop AI/ML Model
B.1 Select State of The Art (SoTA) AI/ML Algorithm(s) / Ensemble Model
Duration: 2 weeks
Dependencies: A.1
B.2 Establish Input Signals, Source Data channels, and Data Sourcing Strategy
Duration: 2 weeks
Dependencies: A.1
B.3 Build Data Extract-Load-Transform [ETL] data pipeline for data collection, curation, annotation, preprocessing, and augmentation, develop sampling strategy and ML model data store
Duration: 4 weeks
Dependencies: B.2
B.4 Establish baseline ML quality metrics in terms of Precision, Recall, and Confusion Metrics
Duration: 2 weeks
Dependencies: B.3
B.5 Implement data and model version control, automated model testing, and model monitoring framework
Duration: 3 weeks
Dependencies: C.2
B.6 Train and Test ML model output, validate ML output via Human-in-the-loop audit, establish ML model monitoring, re-training, and auto enforcement strategy
Duration: 4weeks
Dependencies: B.3
B.6 Optimize ML model performance through feature engineering and hyperparameter tuning
Duration: 2 weeks
Dependencies: B.5
B.7 End-to-end validation in shadow/proxy mode to establish production baseline and meet non-functional requirements [scalability, performance, security, and response-time requirements]
Duration: 4 weeks
Dependencies: B.5
B.8 Establish infrastructure monitoring and model capacity management procedures
Duration: 4weeks
Dependencies: B.6
B.9 Develop MLOps (CI/CD) operating guidance to reduce model development to deployment time and operational infrastructure cost
Duration: 4 weeks
Dependencies: B.8
B.10 Establish weekly ML model audit process to evaluate false positives [FPs] and False Negative [FNs] and develop a feedback loop based on Audit per Week [APW] capacity
Duration: 2 weeks
Dependencies: B.9
C: Conduct User Testing and implementation Framework
C.1 Conduct User Testing / A/B Testing / False Positive validation
Duration: 3 weeks
Dependencies: B
C.2 Ehance, Modify ML model - Human Audit sampling strategy
Duration: 2 weeks
Dependencies: C.2
C.3 Establish Auto and Manual Enforcement / Model outcome implementation workflow, policy updates, and stakeholder communications/reason codes/error reporting mechanism
Duration: 4 weeks
Dependencies: C.2
C.4 Develop a reporting Key Performance Indicator (KPI) dashboard, progress monitoring metrics
Duration: 3 weeks
Dependencies: C.3
C.5 Automate routine tasks and workflows, reduce manual efforts and human errors
Duration: 3 weeks
Dependencies: C.3
C.6 Product Documentation, Training, and system implementation workflows
Duration: 3 weeks
Dependencies: C.3
D: Integrate User Feedback and develop ML model integration strategy
D.1 Integrate User Feedback, establish internal and external request submission workflow
Duration: 2 weeks
Dependencies: B, C
D.2 Establish Service Level Agreements to address customer-facing issues/escalations/remediation procedures
Duration: 2 weeks
Dependencies: B, C
D.3 Develop API-based model integration to support 3rd party integration efforts
Duration: 6 weeks
Dependencies: B, C
D.4 Document out-of-scope use cases with supporting rational
Duration: 4 weeks
Dependencies: B, C
D.5 Publish resource allocation, bandwidth utilization metrics/dashboard
Duration: 3 weeks
Dependencies: B, C
D.6 Explore cross-team collaboration opportunities
Duration: X weeks
Dependencies: B, C
D.7 Benchmark Product Performance against best-in-class models/industry standards
Duration: 3 weeks
Dependencies: B, C
E: Launch ReviewShield
E.1: Launch ReviewShield
Duration : 3 Weeks
Dependencies: B, C
E.1: Launch ReviewShield
Duration : 3 Weeks
Dependencies: B, C
E.1: Launch ReviewShield
Duration : 3 Weeks
Dependencies: B, C
E.1: Launch ReviewShield
Duration : 3 Weeks
Dependencies: B, C
Definition: CPM is a project management technique used to identify the longest sequence of dependent tasks (the critical path) that determines the shortest time in which a project can be completed. By focusing on these critical tasks, product managers can effectively allocate resources and manage time to ensure project deadlines are met.
In an AON diagram, each node (or circle) represents an activity, and the arrows between them represent dependencies. This format allows for a clear visualization of task sequences and dependencies.
Resource Allocation: CPM allows project managers to focus on critical tasks, ensuring that resources are allocated effectively to avoid delays.
Risk Management: By identifying the critical path, teams can monitor these tasks closely to mitigate risks that could lead to project overruns.
Timeline Planning: CPM provides a clear timeline for project completion, helping stakeholders understand project milestones and deadlines.
Efficiency: By prioritizing critical tasks, teams can improve efficiency and ensure that key objectives are met without unnecessary delays.