Skip to main content
Devansh Singh
Systems & Case Studies

Devansh Singh Projects

A documented archive of software systems, distributed architectures, and machine learning prototypes. Each project includes an architectural analysis, real trade-offs, and lessons learned.

StockBeacon

High-Frequency Algorithmic Stock Signal Engine

A low-latency predictive pipeline for stock market volatility and direction signals using order book imbalances and multi-layered LSTM networks.

PythonTensorFlow / KerasReactNode.jsWebSocketsRedisDocker
Case Study AvailableCode

Chatty

End-to-End Encrypted Peer-to-Peer Messenger

A private, zero-knowledge WebRTC messaging client that avoids centralized database logging by exchanging ephemeral cryptographic keys directly in browser sandboxes.

ReactWebRTCSocket.ioWeb Crypto APITailwind CSSTypeScript
Case Study AvailableCode

AI Bot

Agentic LLM Workspace Controller & Background Worker

An autonomous background worker that hooks directly into terminal shells to coordinate local git operations, directory structures, and scheduled routines via semantic tool calls.

Node.jsTypeScriptGemini APIShellJSZsh Hooks
Case Study AvailableCode

MCP Server Agent

Autonomous Model Context Protocol Server for Systems Automation

A production-grade TypeScript implementation of Anthropic's Model Context Protocol (MCP), providing AI clients with sandboxed system inspection and execution capabilities.

TypeScriptNode.jsModel Context Protocol SDKShellJSZsh
Case Study AvailableCode

Lack of Education: Bihar Edition

Public Demographic & Education Data Visualizer

An open-source interactive data dashboard charting school facilities, literacy imbalances, and rural resource distributions across districts in Bihar.

ReactD3.jsTailwind CSSGeoJSONChart.js
Case Study AvailableCode

Open Source & Reproducibility

I strive to publish source code, reproducible benchmarks, and architecture explanations for my personal engineering experiments. For enterprise projects subject to NDA, only non-confidential architectural patterns and anonymized metrics are discussed.