Go

This curriculum takes you from advanced beginner to intermediate Go developer through five practical projects. Each project builds on the last and introduces patterns you’ll use in production systems.

Project 1: ghstat, GitHub repository CLI

Difficulty: 3 / 5

What you build

ghstat is a CLI tool that fetches and displays statistics for one or more GitHub repositories: stars over time, top contributors, and open issues by label. A local cache prevents redundant API calls on repeated runs.

Requirements

• Implement subcommands (ghstat repo stats, ghstat repo compare) using Cobra.
• Configure the HTTP client (timeout, retry count, auth token) using the functional options pattern.
• Fetch multiple repositories concurrently using goroutines and errgroup.
• Cache responses to disk as JSON. Invalidate the cache after a configurable TTL.
• Wrap all errors with context: fmt.Errorf("fetching repo %s: %w", name, err).

What you learn

• Cobra subcommand structure and flag binding
• Functional options pattern for configuring structs
• Concurrent fan-out with errgroup
• JSON serialization and file I/O for caching
• Error wrapping with %w

Why it matters

CLIs are the fastest path to shipping something real in Go. This project forces a clean package boundary between a github client package and a cmd package. The fan-out pattern you implement here reappears directly in Projects 3 and 5.

Project 2: Webhook ingestion service

Difficulty: 3.5 / 5

What you build

You’ll build an HTTP service that receives, validates, and stores inbound webhooks from any HMAC-signed provider (GitHub, Stripe, and similar). The service exposes three endpoints:

• POST /hooks/:provider: receive and store an event
• GET /events: list stored events, filterable by provider and status
• POST /events/:id/replay: re-deliver a stored event to a configurable target URL

Requirements

• Build the server using net/http directly. Do not use a router framework.
• Implement HMAC-SHA256 signature verification as a func(http.Handler) http.Handler middleware. Read and buffer the request body once. Pass it to downstream handlers without consuming it twice.
• Define a Store interface with Save, List, Get, and MarkReplayed methods. Implement it as an in-memory struct guarded by a sync.RWMutex.
• Inject http.Client as a dependency for the replay handler. Do not use http.DefaultClient.
• Implement graceful shutdown: listen for os.Signal, call server.Shutdown(ctx) with a timeout, and drain in-flight replay requests before exiting.
• Write table-driven tests using httptest.NewRecorder. Include cases that send invalid signatures and confirm rejection.

What you learn

• net/http handler and middleware composition
• HMAC signature verification, the standard webhook security pattern
• Interface-driven storage design
• Graceful shutdown with context.WithTimeout
• Testing HTTP handlers with httptest

Why it matters

Webhook handling appears in nearly every backend integration. Stripe, GitHub, Twilio, and most SaaS platforms use the same HMAC pattern. Injecting http.Client as a dependency rather than using http.DefaultClient is a small habit with large consequences for testability. The Store interface you define here is the same pattern you’ll use in Project 4 with a real database.

Project 3: logwatch, concurrent log file analyzer

Difficulty: 4 / 5

What you build

logwatch ingests one or more log files concurrently, parses lines against configurable patterns (error rate, slow requests above a threshold, IP frequency), and emits a structured summary report to stdout or as JSON.

Requirements

• Implement a worker pool: a fixed number of goroutines consume from a shared chan string of lines. Do not spawn one goroutine per line.
• Structure processing as a pipeline with three stages: producer → parser → aggregator. Connect each stage with a channel.
• Propagate a context.Context through every stage. When the user cancels (Ctrl-C), all stages must drain and exit cleanly. No goroutine leaks.
• Use sync.Pool to reuse line buffers and reduce garbage collection pressure on large files.
• Read files with bufio.Scanner.
• Emit structured output using log/slog (Go 1.21+).

What you learn

• Worker pool pattern with bounded concurrency
• Pipeline pattern: staged processing over channels
• Context cancellation and goroutine lifecycle management
• sync.Pool for buffer reuse
• Structured logging with slog

Why it matters

After completing this project, you’ll have implemented the two concurrency patterns that appear most often in real Go systems: worker pools and pipelines. You’ll find both in data ingestion systems, ETL pipelines, stream processors, and CI runners. Context cancellation is required knowledge for any production Go codebase. Every blocking operation must respect it.

Project 4: Task management API with PostgreSQL

Difficulty: 4 / 5

What you build

You’ll build a REST API for managing tasks, users, and labels, backed by PostgreSQL. Users authenticate with JWT. All data operations run against a real database with versioned migrations.

Requirements

• Use pgx with pgxpool for connection pooling. Set the pool size explicitly. Do not rely on defaults.
• Write database migrations with golang-migrate. Run migrations at startup.
• Implement transactions for multi-step operations (for example, creating a task and assigning labels atomically).
• Define custom error types: ErrNotFound, ErrConflict, and ErrUnauthorized. Map them to HTTP status codes in a single handler helper. No type switch per handler.
• Implement JWT authentication as middleware, building on the middleware chain from Project 2.
• Write integration tests using testcontainers-go: spin up a real PostgreSQL container per test run. Do not mock the database layer.

What you learn

• pgx and pgxpool for production database access
• Schema versioning with golang-migrate
• Transaction handling for atomic multi-step writes
• Custom error types that carry behavior, not just messages
• Integration testing with real infrastructure via testcontainers-go
• JWT middleware layered on an existing middleware chain

Why it matters

Most production Go services talk to a database. Understanding pgxpool prevents connection exhaustion, one of the most common production incidents in Go. Integration tests with testcontainers-go catch real bugs that mock-based tests hide. Custom error types let you handle failures consistently across handlers without duplicating logic.

Project 5: Reverse proxy with health checks and circuit breaking

Difficulty: 4.5 / 5

What you build

You’ll build an HTTP reverse proxy that load-balances across a configurable list of backend addresses. Background health checks remove unhealthy backends from rotation. A per-backend circuit breaker stops forwarding to failing backends and retries them after a cooldown period. The proxy exposes Prometheus metrics on /metrics.

Requirements

• Use httputil.ReverseProxy. Customize it with a Director function and a ModifyResponse hook.
• Run background health checkers on a time.Ticker. Each checker must respect context cancellation and exit cleanly when the proxy shuts down.
• Implement a circuit breaker state machine (closed → open → half-open) as a struct guarded by sync.Mutex. Do not use a third-party circuit breaker library.
• Use sync/atomic for high-frequency counters (in-flight requests, failure counts) to avoid lock contention.
• Instrument the proxy with the Prometheus Go client: expose in-flight request counts, request duration histograms, and a backend health gauge.
• Configure backends, timeouts, and circuit breaker thresholds through functional options.

What you learn

• httputil.ReverseProxy customization
• Background goroutine lifecycle management with context
• Circuit breaker state machine design
• sync/atomic for low-contention counters
• Prometheus instrumentation: counters, histograms, gauges
• Functional options for complex component configuration

Why it matters

A proxy concentrates the hardest production Go problems in one place: concurrent state mutation, goroutine lifecycle, and observability. The circuit breaker is a foundational distributed systems pattern used in virtually every service-to-service call at scale. After this project, you can read and contribute to production codebases like Traefik, Caddy, or internal platform tooling.

Getting Started

Fundamentals

Types and collections

Data Formats

Concurrency

Errors and Panics

Clients

Servers

Storage

Context

Networking

CLI Tools

Templates

Testing

System Programming

Patterns

Telemetry

Shipping