Build projects that mirror real production work: each should include tests, CI, containerization, basic observability, and a short case study with measurable outcomes like latency, error rate, or cost; this combination signals job readiness across roles.
1) Auth-enabled API service
- Implement a REST or GraphQL API with signup/login, role-based access, validation, pagination, rate limits, and consistent error models.
- Add integration tests, OpenAPI docs, Docker, and a free-tier deploy with a rollback plan; report p95 latency and one optimization.
2) Payments or booking workflow
- Simulate idempotent operations with retries, queues, and outbox pattern to ensure consistency under failures.
- Compare indexed vs non-indexed queries and introduce a small cache; publish before/after latency and a data integrity note.
3) Data pipeline: CDC → warehouse → BI
- Ingest mock OLTP changes, transform into a star schema, enforce data-quality tests, and surface a dashboard for 3 KPIs.
- Define SLAs, lineage, and a backfill strategy; include a cost/performance note and a failure recovery runbook.
4) Streaming analytics slice
- Consume events, compute windowed aggregates and anomaly alerts, and expose a minimal API or dashboard.
- Demonstrate exactly-once semantics or acceptable at-least-once behavior; quantify alert precision/recall over a test set.
5) Tabular ML system end-to-end
- Build baseline vs improved model with proper validation, feature importance, and a FastAPI inference endpoint.
- Add drift detection, model card, and an A/B test plan; track accuracy and latency targets with a rollback path.
6) GenAI RAG application
- Ingest domain docs, test chunking/embedding strategies, design prompts, and add a safety filter with refusal cases.
- Create an offline evaluation set for answer quality; track latency/cost per query and document failure modes and fixes.
7) DevOps/SRE platform slice
- Provision a service with IaC, CI/CD, artifact signing, and policy checks; implement blue/green or canary releases.
- Instrument logs/metrics/traces, define one SLO, run a chaos or rollback drill, and publish a brief postmortem.
8) Hardened web app (AppSec)
- Add secrets management, parameterized queries, dependency/SBOM scans, and least-privilege IAM or role permissions.
- Write a short threat model and fix two vulnerabilities (e.g., XSS/CSRF); show CI failing on unsafe changes.
9) Accessible dashboard or mobile app
- Build an accessible, responsive UI or offline-first mobile app with local storage and sync conflict resolution.
- Track Lighthouse scores, FCP/LCP or app startup time; include an a11y checklist and two performance fixes.
10) Team project with PR reviews
- Collaborate on a multi-repo feature using issues, branches, and small PRs with code reviews and a release plan.
- Produce a design doc, ADRs, and a 5-minute demo; quantify team outcomes (bug count reduced, CI time improved, deployment success rate).
How to package each project
- README with setup, endpoints or screens, diagram, and test instructions; CI badge green and one-command run.
- Case study: problem, approach, metrics (before/after), lessons, and next steps; link a 2–4 minute demo video.
- Evidence of reliability and security: logs/dashboards screenshot, SLO summary, secret scanning, and a brief postmortem.
6-week build plan for one flagship
- Weeks 1–2: Core feature set, DB schema, tests, CI, Docker; ship v1 and record a quick demo.
- Weeks 3–4: Observability, SLO, security pass (scans, secrets), and a performance improvement with measured impact.
- Weeks 5–6: Deploy with IaC, canary/rollback drill, postmortem, polish docs and case study; request one external code review.
Focusing on these ten projects—with tests, CI/CD, observability, security, and measurable outcomes—creates a compact, production-minded portfolio that stands out for internships and entry-level roles.