Anshuman Singh

B.Tech Computer Science student focused on local-first AI, computer-vision systems, applied machine learning, and polished front-end interfaces.

I turn academic and personal projects into usable software: desktop assistants with local inference, Android release pages for computer-vision apps, reproducible ML notebooks, and portfolio systems that are fast, visual, and open source.

About Me

EducationBachelor of Technology (Computer Science Engineering) Jagannath University · 09/2023 – 06/2027
FocusAI, computer vision, ML, and front-end systems
DirectionAI tools and Machine learning

Projects

ActiveDesktop AI

Jarvis Local AI Assistant

Windows desktop assistant with local LLM inference and voice control.

A Windows-first desktop assistant built with Python and PySide6. It uses Ollama for local reasoning, Whisper for speech-to-text, SQLite for memory, and bounded autonomy rules for safer local actions.

PythonPySide6OllamaWhisperSQLite
Role: Solo build - design, engineering, local AI workflow
Release HubComputer Vision

Smart Driver Monitoring Release Hub

Android APK release page for a computer-vision driver-safety system.

A release and download surface for a Smart Driver Monitoring Dashboard. The project makes an Android computer-vision app easier to distribute, document, and test.

AndroidAPKComputer VisionDriver SafetyRelease Eng
Role: Release engineering and public distribution
Applied MLResearch

Variable Star Classifier

Machine-learning notebook for classifying periodic variable stars.

A Jupyter Notebook project for classifying periodic variable stars from astronomical data. It highlights applied machine learning, feature engineering, and reproducible research workflows.

PythonJupyterscikit-learnAstronomyGaia DR3
Role: Research notebook and applied ML
LiveWeb AI

Humanify

Rewrites AI-sounding text into natural, human writing.

A Next.js app that runs a rewrite pipeline over Gemini models: preflight, diagnosis, fact-lock, rewrite, and final quality check. Built privacy-first, with text processed in-request.

Next.jsTypeScriptGeminiServer ActionsPrivacy-first
Role: Solo build - full-stack design and engineering
CollabDiscord AI

Zinging

Production-oriented Discord assistant bot powered by NVIDIA NIM.

A cog-based Discord assistant with AI ask/code workflows, translation mode, database-backed memory, command controls, caching, and Railway deployment readiness.

PythonDiscordNVIDIA NIMPostgresRailway
Role: Collaboration - AI assistant bot and deployment-ready backend
CollabCreative Web

ArtGridX

Scrapbook-style portfolio system for a drawing artist.

A premium artist portfolio with oversized parallax typography, masonry collage galleries, animated lightbox, Supabase contact flow, and a mobile-friendly admin dashboard.

Next.jsTypeScriptSupabaseTailwindFramer Motion
Role: Collaboration - portfolio interface and content-management workflow

Stack I use to build, test, and ship

Tools behind the public repos: local models, native desktop UI, static front-end delivery, and release packaging.

Languages

  • Python
  • TypeScript
  • JavaScript
  • HTML
  • CSS

AI & ML

  • Ollama
  • Whisper
  • PyTorch
  • scikit-learn
  • Jupyter

App & Web

  • PySide6
  • Next.js
  • React
  • Tailwind CSS
  • Anime.js

Delivery

  • Git & GitHub
  • GitHub Actions
  • Android (APK)
  • Static export

Have a project, role, or collaboration in mind?

I am open to software roles, internships, collaborations, and product engineering work where the implementation has to survive real use.