AlgoVisuals is a learning platform built around a simple belief:

Algorithms click when you can see them run.

We pair the rigor of a competitive-coding judge with the clarity of a great teacher — every problem ships with animated, step-by-step traces of every solution approach, so learners don't just submit code, they understand it.

LeetCode-style platforms test you. Textbooks explain to you. Almost nothing shows you what a heap rebalances into, or how a sliding window actually slides. That gap is where most learners stall. AlgoVisuals closes it.

A polyglot workspace of focused services, each in its own repo:

algovisuals/
├── service/              FastAPI backend — problems, submissions, streaks
├── judge-service/        Sandboxed code execution and grading
├── ui/                   Next.js frontend
├── algovisuals-problems/ Problem definitions, test cases, trace generators
├── algovisuals-deploy/   AWS CDK infra (api gw + lambda + RDS + judge VM)
└── docs/                 Workspace-level documentation

Deployed on AWS (ap-south-1) with API Gateway, Lambda, RDS Postgres, a dedicated judge VM, and GitHub Actions CI/CD via OIDC.

We're heads-down on three tracks for the next quarter:

• Visualizer overhaul — smoother animations, better edge-case rendering, side-by-side compare across approaches.
• Editorial redesign — every optimal solution gets the teach-a-beginner treatment.
• Problem audit — quality pass across the full catalogue.

• Public profiles, shareable trace links, and a lightweight discussion layer.
• Mobile-first reading experience for editorials and traces.

These are the moves we believe will define the next chapter of AlgoVisuals.

A real-time, conversational tutor you can talk to while you solve.

• Think aloud, get nudged. Walk through your approach verbally — the tutor listens, asks clarifying questions, and points out gaps without spoiling the answer.
• Socratic by default. Hints are progressive: a gentle nudge first, then a stronger one, then a worked partial step — never the full solution unless you ask.
• Code-aware. The tutor sees your editor state, your test results, and the problem's trace data, so its guidance is grounded in your code, not generic advice.
• Interview simulation. A "mock interview" mode where the tutor plays the role of a senior engineer — timed, probing, with a written debrief at the end.

The goal: every learner gets a patient, infinitely-available mentor — the kind that, until now, only the luckiest among us had.

Personalized, time-boxed roadmaps that turn "I have an interview in 6 weeks" into a concrete daily plan.

• Target the role. Pick your target company tier and role (frontend, backend, systems, ML) — the plan adapts to the patterns those interviews actually test.
• Calibrated difficulty. A short diagnostic gauges your current level; the plan starts where you are, not at problem #1.
• Pattern-first sequencing. Topics are ordered the way they build on each other — two pointers → sliding window → monotonic deque — not alphabetically.
• Daily rhythm. Each day mixes a new concept, a guided problem, a fresh problem, and spaced-repetition review of past stumbles.
• Progress that means something. Mastery is measured by problems solved without hints on first attempt, not raw submission count.

The goal: replace the anxiety of "what do I study tonight?" with a plan you can trust.

AlgoVisuals is built in the open across this workspace. If you're a learner, an educator, or just curious — explore the repos, file issues, and tell us what's missing.

• Product: algovizuals.com (replace with live URL)
• Issues & feedback: open a GitHub issue on any of the service repos
• Contact: vashiraj2000@gmail.com