# ADR-001: Use Go for API Server

## Status
Accepted

## Context
We needed to choose a programming language for the Fetch ML API server that would provide:
- High performance for ML experiment management
- Strong concurrency support for handling multiple experiments
- Good ecosystem for HTTP APIs and WebSocket connections
- Easy deployment and containerization
- Strong type safety and reliability

## Decision
We chose Go as the primary language for the API server implementation.

## Consequences

### Positive
- Excellent performance with low memory footprint
- Built-in concurrency primitives (goroutines, channels) perfect for parallel ML experiment execution
- Rich ecosystem for HTTP servers, WebSocket, and database drivers
- Static compilation creates single binary deployments
- Strong typing catches many errors at compile time
- Good tooling for testing, benchmarking, and profiling

### Negative
- Steeper learning curve for team members unfamiliar with Go
- Less expressive than dynamic languages for rapid prototyping
- Smaller ecosystem for ML-specific libraries compared to Python

## Options Considered

### Python with FastAPI
**Pros:**
- Rich ML ecosystem (TensorFlow, PyTorch, scikit-learn)
- Easy to learn and write
- Great for data science teams
- FastAPI provides good performance

**Cons:**
- Global Interpreter Lock limits true parallelism
- Higher memory usage
- Slower performance for high-throughput scenarios
- More complex deployment (multiple files, dependencies)

### Node.js with Express
**Pros:**
- Excellent WebSocket support
- Large ecosystem
- Fast development cycle

**Cons:**
- Single-threaded event loop can be limiting
- Not ideal for CPU-intensive ML operations
- Dynamic typing can lead to runtime errors

### Rust
**Pros:**
- Maximum performance and memory safety
- Strong type system
- Growing ecosystem

**Cons:**
- Very steep learning curve
- Longer development time
- Smaller ecosystem for web frameworks

### Java with Spring Boot
**Pros:**
- Mature ecosystem
- Good performance
- Strong typing

**Cons:**
- Higher memory usage
- More verbose syntax
- Slower startup time
- Heavier deployment footprint

## Rationale

Go provides the best balance of performance, concurrency support, and deployment simplicity for our API server needs. The ability to handle many concurrent ML experiments efficiently with goroutines is a key advantage. The single binary deployment model also simplifies our containerization and distribution strategy.