Backend Omniscience.
A backend mastery gauntlet inspired by the roadmap.sh backend roadmap. The app turns a giant topic map into a guided, from-zero curriculum: 1,557 problems, 498 runnable coding drills, interview-answer practice, request-lifecycle explanations, confidence tracking, and capstone projects. Progress is saved locally. No account required.
A roadmap tells you what to learn. It does not teach you to operate it.
The roadmap.sh backend roadmap is useful because it names the terrain: HTTP, APIs, databases, auth, caching, queues, deployment, observability, scaling, and system design. The problem is that a map is not the same as training. A learner can know the next box to study and still not know how to write an endpoint, debug a slow query, explain OAuth, or defend a design choice in an interview.
Backend Omniscience is built for that missing layer. It assumes the user starts with no backend context, then moves from definitions to guided examples, checks, coding drills, debugging prompts, interview answers, and projects. The point is not to clone roadmap.sh. The point is to turn the roadmap into reps.
The current live build is closer to a bootcamp workbook than a checklist: 460 hours of curriculum across fundamentals, JavaScript, Python, HTTP, API design, auth, SQL, performance, architecture, DevOps, distributed systems, framework tracks, projects, and capstones.
Three choices worth defending.
The ones with real tradeoffs.
The curriculum follows a fundamentals-to-system-design spine.
The course is ordered so each subject builds on the previous one: programming fundamentals first, then Internet and HTTP, API design, auth, SQL, caching and performance, queues, object storage, observability, deployment, scaling, distributed systems, system design, language/framework tracks, projects, and capstones.
That was the main product challenge: not just adding content, but making the sequence feel teachable. Within each subject, problems are sorted into a learn-then-defend arc: glossary, lesson, quiz, coding, debug or design prompt, project, and oral exam.
Coding drills run in the browser, but the app is still a static deploy.
A lot of the curriculum is not just reading or multiple choice. JavaScript, TypeScript, Python, Node, Flask, and Django tracks include runnable assessments for small functions, API-shaped logic, parsing, validation, caching, retry behavior, and framework patterns.
The grading layer is intentionally isolated from the UI. Test specs live alongside the curriculum data, execution happens off the main interface path, and progress stays local. That keeps the live site easy to host on Vercel while still feeling like a real practice tool.
Interview fluency is treated as a first-class feature.
The app includes “Explain it in an interview” answers at three levels: simple, senior, and system-design. It also has a “follow the request” view that walks a request from client to load balancer, routing, auth, validation, service logic, database transaction, queue, logs, response, and retry/idempotency failure modes.
That layer is the product’s wedge: roadmap.sh tells you the topics; Backend Omniscience trains you to sound like someone who has built, debugged, and operated those topics.
A large, living backend workbook.
As of July 2026, the live snapshot has 1,557 total problems, 498 runnable coding drills, 58 interview-answer triads, a 324-term encyclopedia, concept-specific quick writes, request-lifecycle cards, project prompts, and a confidence ladder from “Not started” to “Can defend.”
The project is still evolving, but it is already useful as a structured self-study track: start from zero, work through the spine, practice code after the concept, then connect subjects through capstones like URL shorteners, webhook receivers, queue-backed email services, and file upload pipelines.
The long-term direction is more depth where it matters most: better problem-specific lessons, richer coding assessments, more interview defense prompts, and tighter progression from fundamentals into production-grade backend design.