Internal Documentation: Language Selection Strategy

Project: Language Governance Integration

1. Executive Strategy Statement

Objective: Provide developers with a clear decision framework for selecting between C# and Python for new services and prototypes.

The organization utilizes a Dual-Core Architecture. We do not ban languages; we optimize for their inherent strengths.

C# is the backbone for Azure-native infrastructure, high-concurrency APIs, and long-term maintainability.
Python is the engine for data science, AI/ML experimentation, and rapid logic discovery.

2. Decision Support Matrix

This table serves as the primary visual guide for developers during the project initiation phase.

Feature / Requirement	Use C# (.NET)	Use Python
Azure Native Integration	Primary Choice (Best-in-class SDKs)	Secondary (Functional, but higher overhead)
AI / Machine Learning	Interface only	Model Training & Logic
Processing Speed	High-throughput / Low-latency	Prototyping / Batch Processing
Type Safety	Strong (Compile-time)	Dynamic (Runtime)
Lifecycle	Multi-year enterprise support	Rapid iteration / "One-offs"

3. "Fast One-Off" & Prototyping Workflow

To prevent architectural "bottlenecks," we define the following expedited path for prototypes:

The "Sandbox" Rule

Developers are authorized to prototype in either language regardless of the final production destination, provided they follow these constraints:

Isolation: Use Docker containers or Azure Functions to keep prototype dependencies from polluting the core environment.
Contract-First: Define the input/output (JSON/OpenAPI) before coding. This allows the prototype to be refactored for production-ready code without breaking the backend.
Time-Boxing: Prototyping phases must be strictly time-boxed; a prototype should not be allowed to live for an indefinite amount of time.

4. Developer Skill-Base Checklist

This interactive checklist ensures that the choice of language aligns with the available human capital and project risks.

Skill Density Does the assigned team have at least one Senior-level developer in the chosen language?

Dependency Audit Does the project require a library that only exists in one ecosystem (e.g., PyTorch for Python or Entity Framework for C#)?

Maintenance Plan If a "one-off" Python script becomes a "permanent" tool, is there a plan to support its environment (e.g., virtualenv, dependency patching)?

Security Baseline Can the project meet the mandatory Azure Identity (Entra ID) requirements using the selected stack?

5. Architectural Standards for Coexistence

To ensure "both" work as "one," the following technical requirements apply:

Interoperability: All services must communicate via REST or gRPC. Direct library sharing between C# and Python is prohibited to avoid tightly coupled "DLL Hell."
Unified Logging: Both languages must use the OpenTelemetry SDK to pipe logs into the existing Azure Log Analytics workspace.
Deployment: All code—whether a 50-line Python script or a 50,000-line C# API—must be deployed via the standard Azure DevOps/GitHub Actions pipeline.