Content ID Platform

A simplified content identification platform inspired by high-volume automatic content recognition (ACR) systems. A client submits media metadata and a fingerprint hash; the .NET API persists it and enqueues a job; a Go worker consumes the job, runs fingerprint matching against a reference catalog, and writes results back.

Architecture

flowchart LR
  Client --> API[ContentId.Api\n.NET 8]
  API --> SQL[(SQL Server)]
  API --> Mongo[(MongoDB)]
  API --> SQS[SQS Job Queue]
  SQS --> Worker[Go Match Worker]
  Worker --> Mongo
  Worker --> SQL
  SQS --> DLQ[SQS DLQ]
  API --> OTel[OTel Collector]
  Worker --> OTel
  OTel --> Tempo[(Tempo\nTraces)]
  OTel --> Loki[(Loki\nLogs)]
  OTel --> Prom[(Prometheus\nMetrics)]
  Tempo --> Grafana
  Loki --> Grafana
  Prom --> Grafana

Loading

Screenshots

Swagger API

Match Result

Grafana — Prometheus Metrics

Grafana — Tempo Traces

Grafana — Loki Logs

Local Quickstart

docker compose up --build

In another terminal, run the end-to-end validation:

python3 scripts/healthcheck.py

Submit a known matching fingerprint manually:

curl -s -X POST http://localhost:18080/v1/submissions \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Interview Demo Track",
    "sourcePlatform": "PartnerUpload",
    "contentType": "audio",
    "durationSeconds": 191,
    "fingerprintHash": "abc123"
  }'

Endpoints

Endpoint	Description
POST /v1/submissions	Submit a fingerprint for identification
GET /v1/submissions/{id}	Get submission status
GET /v1/submissions/{id}/matches	Get match results
GET /health	Dependency health check (SQL Server, MongoDB, asset catalog)
GET /metrics	Prometheus metrics stub

Observability

All services export OpenTelemetry signals to the OTel Collector, which fans out to:

Signal	Backend	Grafana query
Traces	Tempo	{} in Explore → Tempo
Logs	Loki	{service_name="content-id-api"} or content-match-worker-go
Metrics	Prometheus	worker_jobs_processed_total, contentid_submissions_created_total

Grafana UI: http://localhost:13000 (no login required)

Key metrics:

• worker_jobs_processed_total{result="matched|no_match|error"} — worker throughput by outcome
• contentid_submissions_created_total — API intake rate
• http_server_request_duration_seconds — ASP.NET Core HTTP latency

Port Map

Service	External port	Notes
ContentId.Api	18080	Swagger at /swagger
SQL Server	11433
MongoDB	27018
LocalStack	4566	SQS, SNS, S3
OTel Collector gRPC	14317
OTel Collector HTTP	14318
Grafana	13000

What This Demonstrates

• C#/.NET 8 minimal API for submission intake and result retrieval with OpenAPI
• Go worker for async content matching via shared-prefix fingerprint similarity
• SQS + DLQ for retryable background processing (up to 3× retry, then dead-letter)
• SQL Server for normalized job state (submissions, match_results with MERGE upsert)
• MongoDB for flexible fingerprint and raw match documents
• LocalStack for local AWS SQS, SNS, and S3 without cloud costs
• OpenTelemetry traces, logs, and metrics from both services via OTLP gRPC
• Grafana + Tempo + Loki + Prometheus full observability stack
• OpenTofu AWS infrastructure definitions in infra/opentofu
• Python end-to-end healthcheck automation in scripts/healthcheck.py

Docs

• Architecture
• Runbook
• JD Mapping
• Interview Talking Points