BackInteractive Game & AI Inference Platform
#14★ FEATUREDLive on Vercel

IPLMind (IPL Genius)

Akinator-style neural predictor guessing IPL cricket legends through dynamic entropy reduction

AIWebGame
LIVECODEMAPARCH MAP
IPLMind (IPL Genius)
ENGINEERING // ARCHITECTURE

How It Connects

IPLMind combines a 15-year IPL historical dataset (spanning player statistics, bowling economies, strike rates, tournament franchises, and auction records) with an adaptive decision-tree entropy algorithm. As the user responds to binary/scalar questions ("Has your player captained an IPL franchise?", "Is he a left-arm spinner?"), the engine dynamically recalculates Information Gain across the remaining candidate pool, asking the optimal discriminative question next. The frontend is built with Next.js, featuring animated state machines, audio feedback, and celebration reveals.

IPLMind (IPL Genius) / ARCH MAP
ARCHITECTURE ENGINE
INITIALIZING REPO...
IPLMind (IPL Genius) architecture diagram
SCROLL TO PANMERMAID.INK
GitDiagramGitHub

System Components

EXECUTION PIPELINE

How It Works

01
User

Game Initialization

User mentally picks an IPL player and begins the guessing sequence.

→ Initial candidate space (~300+ IPL players)
02
Decision Engine

Entropy Evaluation

Calculates Information Gain for all unanswered questions across remaining candidates.

→ Optimal question with highest partition entropy
03
User Input

User Feedback Evaluation

User selects "Yes", "No", "Don't Know", or "Probably".

→ Scalar confidence modifier
04
Algorithmic Core

Candidate Pool Reduction

Multiplies player probabilities by likelihood factor and renormalizes vector.

→ Pruned candidate list and top certainty percentage
05
Client UI

Prediction Reveal

Presents dynamic player card with photo, IPL team badge, and career achievements.

→ Guessed Player Verification
STACK // ARCHITECTURE INTEL⚡ LIVE SPECS

Technical Breakdown

Granular architectural layers, runtime dependencies, and audited production decisions.

  • Next.js with responsive Tailwind styling
    copy
  • Framer Motion spring physics for question transitions and confidence meters
    copy
  • Sound synthesis using Web Audio API for arcade-like feedback
    copy
  • Client-side execution for instant 0ms question transitions with no server lag
    copy
  • Vercel Edge runtime for static payload distribution
    copy

Engineering Tradeoffs & Key Decisions

Why this architecture over alternatives
2 ARCHITECTURAL CHOICES
DECISION #1AUDITED

Client-Side Decision Tree over Server Roundtrips

↳

Executed the entire entropy algorithm in the browser rather than on an API server.

Delivers instantaneous response times (0ms latency between questions) and zero server hosting costs.

PRODUCTION TESTEDZERO REGRESSIONS
DECISION #2AUDITED

Fuzzy Bayesian Confidence over Strict Boolean Pruning

↳

Allowed probabilistic weights rather than outright eliminating candidates on negative answers.

Prevents game failure when users make factual errors about a player debut year or minor team history.

PRODUCTION TESTEDZERO REGRESSIONS