samarth@fedora:~$ whoami
Backend Developer | DevOps Engineer | System Administrator | Full Stack Architect
samarth@fedora:~$ cat location.txt
Indore, India | 3rd Year CS Student | Available for Internships & freelancing 

about • experience • projects • research • tech stack • fun stuff

Started with Harvard's CS50, then spent the last few years exploring operating systems, network protocols, distributed systems, and now diving deep into WebAssembly, Local LLMs, and API design.

I learn through building. Currently:

• Working: Backend Engineering Intern @ Brand Collabs working on cashcard /— building security pipelines, deploying containerized infrastructure, auditing APIs
• Exploring: WebAssembly/FFmpeg, Ollama (local LLMs), real codebase architecture
• Building: WASM video editor, Local LLM interfaces, research on production systems

📍 Based in: Indore, India
🎓 Education: SAGE University Indore (B.Tech Computer Science, Expected 2026) · CGPA 7.8/10
📧 samarth07nagar@gmail.com | 💼 LinkedIn | 🐙 GitHub | 🌐 Portfolio

Backend Engineering Intern @ Brand Collabs (Jan 2026 – Present) dinehub

• Built static code generation pipeline automating asset workflows, eliminating manual steps
• Deployed fully offline Supabase instance using containerized infrastructure for local testing
• Audited API security across internal services, identified and resolved critical vulnerabilities
• Authored backend documentation and technical workflows, reducing onboarding time

Smart India Hackathon Finalist (2025)
Built healthcare mapping platform for locating nearby medical facilities, competing nationally.

MSME Incubation Collaboration
Designed data collection pipelines and system architecture with postgraduate researchers (government-backed initiative).

WASM Video Editor — Browser-Based Video Processing
WebAssembly, FFmpeg, Node.js · View on GitHub
Built fully client-side video editor using WASM + FFmpeg. Real-time encoding, decoding, export pipelines — desktop-grade media processing inside the browser. Zero server compute.

Local LLM Interface — Streaming Chat for Offline AI
Node.js, Ollama, Streaming APIs · View on GitHub
Developed streaming chat interface for locally-hosted AI models. Backend APIs for model lifecycle management (load, unload, swap) — reduced cold-start latency and improved runtime orchestration.

Personal Blog Platform — SEO-Optimized Static Site
Next.js, MDX · Live
Built SEO-optimized static blog with server-side rendering and edge caching. Sub-200ms page load times on Vercel's edge network.

react.dev Architecture Deep Dive (2025)
Next.js 15, React 19, TypeScript, Tailwind, MDX
Analyzed real production codebase to understand high-quality architecture decisions.

Key Findings:

• No state management library needed — URL + useState + Context sufficient for most sites
• 50+ custom MDX components behind each doc page
• React Compiler already running in production
• Custom SSG pipeline that serializes React tree to JSON at build time
• Build-time eval() workarounds acknowledged as intentional hacks

📄 Full Report: casestudy-reactdotdev.samarth.page
📝 Detailed Notes: Notion

My main setup and configurations live in my dotfiles repository: dotfiles

It includes configs for Neovim, tmux, shell environments, terminal tooling, and my Linux workflow. Most of my development environment revolves around Fedora Linux, tmux, Alacritty, and terminal-first tooling.

System:

• Moto Book 60 Pro
• Intel Core Ultra 5 225H
• 16GB LPDDR5 RAM
• Fedora Linux 43 (since Fedora 38)
• GNOME 49

Main Workflow:

• Alacritty - terminal
• Neovim - editor
• tmux - multiplexer session manager
• OpenCode - main AI agent
• Pi - side AI agent
• fish - shell

Other tools: Yazi, Lazygit, Lazydocker, btop, Starship, Atuin, Zoxide, hunk