Skip to content

Roadmap

Future improvements, feature ideas, and open questions for the ML Interview Prep Kit. Contributions and discussion welcome — open an issue or submit a PR.

Features & Enhancements

Progress Tracking

  • Add a progress tracker (inspired by mission.html panel stats) — should it be part of mind map @mindmap metadata, a separate config file, or localStorage-based?
  • Handle progress for modules outside the mind map (System Design, Coding Practice, Career Frameworks, etc.)
  • Consider a unified "completion state" data model across all overview prototypes

JD-Specific Resume Tools

  • Resume generator/modifier that takes a job description and tailors content from the knowledge base
  • AI prompt template for JD-to-resume alignment
  • LaTeX resume generator — produce a polished, ATS-friendly PDF resume from knowledge base content and a given JD

Research Papers Section

  • Add a section for reading and summarizing research papers
  • Decide: separate top-level folder (research-papers/) or nested under knowledge-base/?
  • Template for paper summaries (title, key contributions, relevance to interviews, link)

Cross-Section Navigation

  • Enable hyperlinking from one folder's content to another in mind maps, Bento/Skill Tree/Mission overviews
  • Make transitions seamless between mind map nodes and docs pages, and between overview prototypes
  • Unified deep-link scheme (e.g., ?focus=NodeName already works for mind map — extend to other views)

Bug Fixes

Mind Map

  • When switching between Radial and Tree layout, the current depth focus (L1/L2/L3) is not retained — the view resets instead of preserving the active depth level

MkDocs Rendering

  • Fix LaTeX rendering issues with $$ $$ delimiters in MkDocs
  • Audit and fix other rendering edge cases (nested lists, complex tables, admonition nesting)

Branding & Design

Identity

  • Finalize a solid repo name, title, and logo
  • Create a standardized color palette, font system, and design tokens shared across all pages (mind map, docs, prototypes)
  • Ensure consistent theme (light/dark) behavior across all entry points

Architecture & Tooling

Documentation Platform

  • Evaluate Quartz (Obsidian-based) vs. MkDocs Material (currently used)
  • MkDocs: mature plugin ecosystem, Material theme, instant navigation, search
  • Quartz: native Obsidian vault support, graph view, backlinks, wikilinks
  • Key question: does the team use Obsidian for authoring? If so, Quartz reduces friction

Reference Implementations & Competitive Research

  • Study AI Engineering from Scratch for inspiration on:
  • Content structure and topic organization
  • UI/UX patterns and navigation design
  • Codebase architecture and build system
  • Evaluate which patterns could improve this repo
  • Survey similar repos, platforms, websites, and tools in the ML/DS interview prep space
  • Catalog what they do well (content depth, UX, interactivity, community, AI features)
  • Identify gaps and opportunities we can differentiate on
  • Draw inspiration across all aspects: content, design, architecture, engagement model

Architecture & Modularity Review

  • Audit the current repo architecture for modularity, scalability, extensibility, and collaboration-readiness
  • Is the folder structure, build pipeline, and content model set up for easy contribution by others?
  • Do we need formal data models (e.g., structured schemas for topics, progress, user state)?
  • Should we adopt specific architectural principles or frameworks (e.g., component-based design, plugin architecture, monorepo tooling)?
  • Define a target architecture that is future-proof
  • Separation of content, presentation, and logic layers
  • Standardized data formats (JSON/YAML schemas for topics, metadata, progress)
  • Clear extension points for new sections, tools, and integrations

AI Integration Strategy

  • Decide the core AI delivery model — this is a fundamental design question:
  • Local/agent mode: Users clone the repo and use it with their own AI agents (Claude Code, Copilot, etc.) locally. Simpler to build, no hosting costs, full user control.
  • Web-hosted mode: AI features embedded in the deployed website (chat, JD analysis, resume gen, adaptive quizzing). Requires hosting, backend infra, database, auth, and cost management.
  • Hybrid: Static site for content + optional local AI workflows (prompt templates, agent configs) that users run on their own machines. Avoids hosting complexity while still being AI-native.
  • Evaluate trade-offs for each model:
  • Web: editing, persistence, scalability, hosting costs, database needs, user auth
  • Local: onboarding friction, dependency on user's local setup, harder to provide seamless UX
  • Hybrid: best of both? Or worst of both?
  • If web-hosted AI features are pursued, scope the infra requirements:
  • Backend (serverless functions vs. dedicated server)
  • Database (user state, progress, generated content)
  • API key management and cost controls
  • Scalability and rate limiting

Overview Prototypes

  • Pick a winning prototype (Bento Grid, Skill Tree, or Mission Control) or combine elements
  • Integrate winner into the build system and deploy to interview.prasanth.io/overview/
  • Add "Overview" nav link to mind map top bar

Last updated: 2026-03-29