Backend Engineering Portfolio Demo

FastAPI in front. Go under load.

A browser-accessible demonstration of typed API validation, service-to-service communication, bounded concurrency, deterministic AI-style enrichment and production-oriented testing.

Live demo online

Mock AI mode is enabled for the hosted demo, making results deterministic and keeping the public environment free of paid API credentials.

Architecture

One request, two runtimes, bounded work.

The public request enters through FastAPI, is validated with Pydantic, then crosses an explicit HTTP service boundary into a Go worker pool before enrichment results are returned in stable input order.

01Browser / API clientJSON batch request
02 · PythonFastAPI gatewayPydantic validation · HTTPX · timeout/error mapping
03 · GoBounded worker poolGoroutines · channels · per-job deadlines · ordered results
04Enrichment providerDeterministic mock or OpenAI-compatible adapter
Interactive demo

Run the pipeline in the browser.

This sends three products through the real hosted FastAPI endpoint and internal Go worker. The response below is rendered from the live API result, not hard-coded UI data.

Request batch3 products
Trail Boots#101

Waterproof hiking boots for rough terrain.

Laptop#102

Portable computer for software development.

Jacket#103

Lightweight outdoor jacket for changing weather.

FastAPI → Go → mock enrichment
Live responseReady
Processed
Failed
Processing
Click Run sample request to exercise the live backend and render the worker results here.
Engineering signals

What this project is meant to demonstrate.

Typed API boundaryFastAPI and Pydantic normalize and reject invalid requests before downstream work begins.
Explicit failure handlingConnection and worker failures are mapped to deliberate 503 and 502 responses.
Concurrency with limitsThe Go service uses a bounded worker pool rather than unbounded goroutine creation.
VerificationPython tests, Go vet/race tests, Docker builds and containerized end-to-end CI protect the architecture.