Devansh Kumar

Building systems that think — from risk intelligence engines to mobile apps. I write code that solves real problems, reads well, and scales.

Hire me
Devansh Kumar Avatar

About Me

I build scalable backend systems, risk intelligence engines, and full-stack web applications with clean architecture.

Engineering Philosophy

  • Decisions matter more than predictions
  • Architecture before implementation
  • Design for failure, not ideal cases
  • Clarity over cleverness
  • Measure trade-offs explicitly
  • Systems over scripts

Systems I Build

Risk EnginesREST APIsML PipelinesAuth SystemsData DashboardsMobile AppsETL PipelinesFraud Detection

Proof of Thinking

    Risk-aware fraud architecture

    Cost-sensitive modeling

    Consequence-weighted thresholding

    EDA-driven reasoning

    Layered decision architecture

    The Extremity Principle

    SIMPLY UNIVERSE - published book

    IEEE Chairperson - IEEE Student Branch Bennett University

Tech Stack

PythonPython
TypeScriptTypeScript
JavaScriptJavaScript
C++C++
CC
JavaJava
DartDart
SQLSQL
FastAPIFastAPI
Next.jsNext.js
ReactReact
Node.jsNode.js
FlutterFlutter
Tailwind CSSTailwind CSS
PythonPython
TypeScriptTypeScript
JavaScriptJavaScript
C++C++
CC
JavaJava
DartDart
SQLSQL
FastAPIFastAPI
Next.jsNext.js
ReactReact
Node.jsNode.js
FlutterFlutter
Tailwind CSSTailwind CSS
PostgreSQLPostgreSQL
RedisRedis
SQLiteSQLite
Scikit-learnScikit-learn
PandasPandas
NumPyNumPy
MatplotlibMatplotlib
DockerDocker
GitGit
GitHub ActionsGitHub Actions
LinuxLinux
REST APIsREST APIs
RailwayRailway
VercelVercel
PostgreSQLPostgreSQL
RedisRedis
SQLiteSQLite
Scikit-learnScikit-learn
PandasPandas
NumPyNumPy
MatplotlibMatplotlib
DockerDocker
GitGit
GitHub ActionsGitHub Actions
LinuxLinux
REST APIsREST APIs
RailwayRailway
VercelVercel

Latest Projects

🖥️

Risk-Aware Fraud Decision System

🛡️

Risk-Aware Fraud Decision System

A multi-axis risk intelligence engine that evaluates financial transactions using behavioral, network, and temporal risk signals to produce explainable fraud decisions.

  • Multi-layered risk scoring with 12+ risk signals
  • Explainable AI — every decision has a breakdown
  • Real-time decision API with sub-100ms latency
  • Adaptive rules engine with manual override support
PythonFastAPIScikit-learnPostgreSQLRedisRailway
🖥️

Flutter OTT Streaming App

🎬

Flutter OTT Streaming App

A Flutter-based OTT streaming-style application with local authentication, content rentals, and a rich cinematic UI.

  • Local auth with secure session management
  • Content catalog with rental & purchase flow
  • Rich cinematic UI with smooth page transitions
  • Cross-platform: iOS, Android & Web
FlutterDartSQLiteBLoC
🖥️

Portfolio — This Site

🌐

Portfolio — This Site

A clean, minimal, and high-performance portfolio website with particle animations, bento grid layout, and GitHub stats integration.

  • Particle canvas background in hero section
  • Bento grid about section
  • Tech stack infinite marquee
  • GitHub stats integration
Next.jsTypeScriptTailwind CSSFramer Motion

GitHub Stats

A snapshot of my open-source activity.

Overview

📁6Public Repos
👥1Followers
➡️2Following
📅2024Member Since

Top Languages

TypeScript
38%
Python
28%
Dart
18%
C++
10%
HTML
6%

Contribution Activity

Devansh's GitHub contribution chart

Recent Blog

Coming soon
🛡️

Building a Risk-Aware Fraud Decision Engine

How I designed a multi-axis scoring system that explains every decision it makes — and why explainability matters more than accuracy.

Coming soon
💙

Why Flutter is Underrated for Rapid Prototyping

From zero to OTT app in 48 hours — how Flutter's widget system and hot reload change the way you think about building UIs.

Coming soon
🧠

INTJ Notes: On Systems Thinking in Software

A reflection on how personality shapes engineering philosophy — and why I obsess over architecture before writing a single line of code.

Coming soon