Also known online as: yashbytes
I build AI-powered products and the infrastructure they run on β an AI-first engineer shipping production systems, from LLMs to Linux.
Currently working on production AI features across live products. Final-year computer science student.
Currently exploring AI agent architectures, distributed systems, Linux internals, and networking.
- π§ Daily driving Arch Linux + Hyprland
- π€ Building AI systems: RAG pipelines, AI agents with tool use, LLM integrations
- π Exploring networking, DNS infrastructure, and self-hosting
- π Building with Python, TypeScript, FastAPI, Next.js, PostgreSQL, Docker, Bash, and C
- β‘ Interested in AI engineering, platform engineering, and distributed systems
AI Engineering & LLM Systems
Software Development
Linux Internals
Networking & DNS
Infrastructure Automation
Containers & Deployment- π€ AI Bhai
PRIVATEβ mobile-first consumer AI assistant. Chat Mode + task-oriented Work Mode, Hindi / Marathi / English + voice, MCP plugin ecosystem - π AtlasRAG β local-first enterprise RAG over PDF / TXT / Markdown with grounded answers and source citations
- π¬ ChatAPI β document-grounded AI agents for local businesses. Embeddable chat widget, JWT auth, dashboards
- π DesignMCP β MCP server giving AI coding agents UI inspiration. Query parsing, TTL cache, stdio transport
- π Distributed T5 fine-tuning β built a compute cluster from university lab computers; fine-tuned T5-base for grammar correction
- π MYcShell β custom C shell for Unix + Windows. Systems programming, start to finish
- π my-own-internet β alt-root DNS infrastructure with custom TLDs. Real networking, not a toy
Software Development
AI Engineering & Agents
Linux
Networking & DNS
Infrastructure Engineering
DevOps & Platform Engineering
Distributed Systems
Systems Programming
Self-Hosting
Open Source- Website: https://yashpatil.dev
- LinkedIn: https://linkedin.com/in/yashbytes
- GitHub: https://github.com/Real-yash
- X: https://x.com/yashbytes
- Email: yashpatil.cse@gmail.com
I enjoy understanding how things work under the hood β
from LLM tool-calling loops to Linux kernels and networking stacks.


