2. Architecture Constraints
Technical Constraints
| Constraint | Description | Impact |
|---|---|---|
| AI Model Latency | Sentiment analysis must complete within 30 seconds | Influences choice of model architecture and caching strategy |
| Data Privacy | Property and borrower data subject to GLBA regulations | Requires encryption at rest and in transit, audit trails |
| Regulatory Compliance | Valuations must be defensible to federal banking regulators | AI decisions must be explainable, not black-box |
| API Integration | Must integrate with existing lender platforms via REST APIs | Standardized data formats, backward compatibility requirements |
| Browser Support | Web interface must support modern browsers (Chrome, Firefox, Safari, Edge) | No IE11 support required |
Organizational Constraints
| Constraint | Description | Impact |
|---|---|---|
| Team Size | 2-person founding team (technical + business) | Emphasis on automation, managed services over custom infrastructure |
| Budget | Seed-stage startup funding | Cloud-native architecture, pay-as-you-go services, no on-prem hardware |
| Time to Market | MVP required within 6 months for pilot lender | Iterative development, core features first, advanced features later |
| Expertise | Deep CRE domain knowledge, moderate ML experience | Partner with AI consultants for initial model architecture |
Legal Constraints
| Constraint | Description | Impact |
|---|---|---|
| GLBA Compliance | Gramm-Leach-Bliley Act requires financial data protection | SOC 2 Type II certification required, access controls, encryption |
| Fair Lending Laws | Valuations cannot exhibit bias based on protected classes | Bias detection in sentiment scoring, regular fairness audits |
| Data Retention | Financial records retention for 7 years (FINRA) | Immutable audit logs, long-term archival storage |
| Export Controls | AI technology subject to export restrictions | Geo-fencing for international customers, compliance reviews |
Conventions
| Convention | Description | Rationale |
|---|---|---|
| Documentation Standard | Arc42 template for architecture documentation | Industry-recognized format, investor familiarity, comprehensive coverage |
| Commit Messages | Conventional Commits specification | Automated changelog generation, semantic versioning alignment |
| Code Style | ESLint + Prettier for TypeScript/JavaScript | Consistent code quality, automated formatting, reduced review friction |
| Diagram Format | Mermaid.js (code-based diagrams) | Version control friendly, easy updates, no proprietary tools |
| API Design | RESTful APIs with OpenAPI 3.0 specification | Standard tooling, auto-generated documentation, client library generation |
These constraints shaped the technology decisions documented in Solution Strategy.