You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
DevHire Cloud is a Java, systems, and DevOps learning project for a recruitment platform: Java 21, Spring Boot 4.0, Next.js, Kafka, OpenSearch, PostgreSQL, Redis, Docker, Kubernetes, Terraform, and a controlled Claude Haiku AI assistant. The repository documents service boundaries, owned data, event reliability, observability, security, CI/CD, and cloud-readiness evidence so reviewers can verify each claim against the code and supporting records.
30-Second Reviewer Brief
Question
Answer
What is being demonstrated?
A microservices hiring platform with candidate, employer, admin/ops, platform, AI, and code-assessment workflows.
The v0.6 UI follows Stitch project projects/5421325194779586117 with a hybrid design system. Public candidate and job-discovery pages use an ITViec-inspired marketplace pattern: search-first red/white surfaces, salary/location/company prominence, compact filters, and mobile-safe job cards without copying ITViec brand assets. Employer, admin, and platform pages keep the "DevHire Cloud Operations" control-plane style: dark navigation, light operational workspace, dense panels, 8px radius, Inter typography, and evidence-heavy status language. Route-matrix screenshots are checked for broken assets, overflow, raw UUIDs, UNKNOWN, loading-only states, fallback banners, smoke labels, mojibake, and hidden assessment payloads.
Client and Assessment UX
Surface
Professional contract
Job marketplace
/, /jobs, and /jobs/[id] prioritize keyword search, city/skill/salary filters, company credibility, salary clarity, login-safe apply states, and mobile layouts.
Candidate assessment
/candidate/assessments presents Cloud Architecture Challenge as a Java-first LeetCode-style studio with CandidateSolution.java, visible/custom runs, locked submission history, and hidden-test redaction.
Employer review
/employer and /employer/applications show assignment/review context, candidate code, runtime/static score, visible and hidden aggregate evidence, risk flags, and immutable raw score.
Admin/Ops
/admin and /admin/monitoring show runner health, fail-closed posture, queue depth, verdict rates, policy/timeout/unavailable rates, and backlog signals.
Cloud State Matrix
Target
State
Verification
Docker Compose
Full local stack for backend, frontend, data, messaging, and observability
docker compose config --quiet
Raw Kubernetes
Renderable manifests, no latest, includes ai-service
kubectl kustomize deploy/k8s
Helm
Local, staging, production, and AWS values
.\scripts\cloud-verify.ps1
Terraform AWS
Apply-ready blueprint validation; no credentials required for CI validation
Release images publish to GHCR as ghcr.io/jasontm17/devhire/<service>:<tag> with commit SHA tags, OCI labels, SBOM, and BuildKit provenance. Docker Hub mirrors are available as docker.io/nguyenson1710/devhire-cloud-<service>:<tag> when the Docker Hub secrets are configured. The current preview set was also pushed locally through Docker Desktop. See container images.
Production code grading requires DEVHIRE_RUNNER_MODE=judge0 and JUDGE0_BASE_URL; local development can keep deterministic preview mode. Use scripts/code-assessment-smoke.ps1 through the Gateway for assign/run/submit/review coverage, and scripts/judge0-smoke.ps1 against assessment-runner-service to verify health, accepted, wrong-answer, compile-error, timeout, and policy-blocked Java submissions. Operational triage lives in the code assessment runner runbook.
Run and Verify Locally
Port convention:
Mode
Gateway
Frontend
When to use
Compose default
8080
3001
Fresh local stack with no port conflicts
Docker Desktop high-port smoke
18080
13001
CI parity, existing local services, or reviewer machines with common ports occupied
v1.0.0 is not released. The v1 roadmap focuses on product UX depth, backend integration maturity, API/event compatibility, observability SLO maturity, deterministic data depth, cloud apply evidence, and supply-chain hardening. See v1 reviewer guide, v1 demo script, and v1 production gap register.
Honest Scope
DevHire Cloud is a learning project that practices production-engineering patterns, not a claim of live customer traffic. It does not claim external penetration testing, a production AWS account, or real customer data. Secrets are not committed; cloud apply remains a separate, credentialed operation.
About
Java 21 Spring Boot microservices learning platform for recruitment with Kafka, OpenSearch, Docker, Kubernetes, Terraform, observability, CI/CD, and a RAG assistant.