Skip to content
View SrabanMondal's full-sized avatar
💭
Finding the "THE ONE"
💭
Finding the "THE ONE"
3 followers · 4 following

Achievements

Achievement: Pull Sharkx2Achievement: YOLOAchievement: Quickdraw

Achievements

Achievement: Pull Sharkx2Achievement: YOLOAchievement: Quickdraw

Highlights

  • Pro

Block or report SrabanMondal

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SrabanMondal/README.md

💫 About Me:

System Programmer | Backend Architect | AI Infrastructure Explorer

I design and scale backend platforms with a focus on clean architecture, industry best practices, and high-performance systems.
For me, perfection is elegance, and elegance is true beauty — I strive to build systems that are not only fast and reliable, but also graceful in design.

What I Build

  • C++ Web Server from Scratch → Learned the internals of multi-threaded IO, full HTTP parsing, and adaptive rate limiting.
  • Custom UDP VPN (SOCKS5-based) → Built a custom UDP transport with session multiplexing, in-place AEAD encryption, and out-of-order packet handling to tunnel TCP traffic through restrictive networks.
  • AI Infrastructure & RAG Systems → Experimental LLM infrastructure beyond demos.
    • retrieval as a policy: query decomposition, routing, and weighting
    • hybrid neuro-symbolic knowledge access
    • ablation, failure-mode analysis, and evaluation tooling
    • Episodic buffers, symbolic daily tables, and chapter-based reasoning for long-term context.
    • Lightweight ADAS perception-and-control pipeline using YOLOPv2 segmentation/detection and MPC steering, optimized for real-time performance on low-power Intel hardware via OpenVINO.
  • Backend & P2P Systems
    • Built a token-based 1v1 P2P chat system over a Kademlia DHT, featuring a fully distributed matchmaking protocol.
    • Developed a secure AES-encrypted MinIO file-storage service with a Google-Drive-like interface, implemented in Go.
    • Implemented a real-time messaging platform optimized with Redis, supporting scalable WebSocket rooms and instant notifications.
  • Local Voice Assistants (V1 → V2 Evolution)
    Progressed from a threaded prototype (Whisper + Coqui + Gemini) to a CPU-first, turn-aware, multiprocessing voice assistant, featuring:
    • event-driven orchestration
    • interruption-safe (barge-in) interaction
    • streaming STT → LLM → TTS pipelines (Moonshine + Silero VAD + Ollama LLM + Piper TTS)
    • fully local inference using ONNX / GGUF models
      This project emphasizes systems design, concurrency, and low-latency control planes, not just model integration.
  • Open-Source: Typeflow → Creator of a full visual, type-safe workflow engine for Python, featuring:
    • drag-and-drop React Flow editor
    • function & class node system
    • DAG compiler + Python orchestrator generation
    • live execution via SSE
      This project pushes toward a future of visual, OOP-driven, AI-native programming.

Philosophy

I enjoy the challenge of scaling elegant systems into production, while still diving into low-level internals out of curiosity.
My passion lies in giving AI a sense of memory and continuity, bridging the gap between stateless models and lifelong companions.

When I’m not coding, I’m usually researching new ideas, sketching system designs, or unwinding with some anime.

“If perfection is sin, I am a sinner — gladly.”

Snake animation

🌐 Socials:

LinkedIn email

💻 Tech Stack:

C C++ Go JavaScript Java HTML5 TypeScript Python Vercel Render Chakra Express.js FastAPI Flask JWT NPM NestJS Next JS NodeJS React React Native Socket.io Three js TailwindCSS Nginx Redis MongoDB MySQL Neo4J SQLite Prisma Postgres Keras Matplotlib mlflow NumPy Pandas TensorFlow PyTorch scikit-learn Scipy Git GitHub FFmpeg Gradle Docker Swagger

📊 GitHub Stats:



🏆 GitHub Trophies

🔝 Top Contributed Repo

Pinned Loading

  1. adaptive-cpp-http-server adaptive-cpp-http-server Public

    A high-performance HTTP server built from scratch in pure C++ (IOCP), with adaptive rate limiting, IP fairness, and overload admission control. Express.js-style API, systems-level internals.

    C++ 1

  2. proxy-vpn proxy-vpn Public

    Proxy-VPN is a high-performance SOCKS5-compatible proxy that tunnels TCP traffic over an encrypted UDP transport. It uses session multiplexing, zero-allocation packet handling, and XChaCha20-Poly13…

    Go

  3. voice-assistant-v2 voice-assistant-v2 Public

    CPU-first, turn-aware local voice assistant with multiprocessing, streaming STT→LLM→TTS, and interruption-safe orchestration.

    Python

  4. NSHG-RAG NSHG-RAG Public

    Research-grade neuro-symbolic RAG framework where retrieval is a policy, not a vector search, built for evaluation, ablation, and reliability analysis.

    Python

  5. Typeflow Typeflow Public

    Typeflow is a visual workflow engine for Python. Design, validate, and execute workflows using type-safe function/class nodes and a React Flow editor. Fully local, modular, and AI-ready.

    HTML

  6. adas-realtime-inference adas-realtime-inference Public

    Lightweight ADAS perception-and-control pipeline powered by YOLOPv2 segmentation, object detection, and MPC steering. Optimized for real-time deployment on low-power Intel hardware using OpenVINO.

    Python