Ishwor K.

I spent years doing backend work in Python and Rust. Somewhere along the way I realized the most valuable thing I could build for a client was not another API endpoint but a system that removes an entire category of manual work from their day. That is what I do now. My stack for automation is n8n and Make. I build multi-step workflows across payment processors, CRMs, databases, and communication tools. Triggered, scheduled, or event-driven, whatever the job needs. On the AI side I integrate LLMs into workflows that actually run in production. I have built RAG pipelines with LangChain backed by pgvector and Supabase, memory-persistent AI agents using Letta, and custom MCP servers with FastMCP so AI tools like Claude Desktop can talk directly to your business systems. Not demos. Deployed systems with real clients. Web scraping is something I have done consistently across five years of client work. Playwright and Selenium for JavaScript-heavy sites, BeautifulSoup and custom parsers for structured extraction, production pipelines with anti-bot handling, pagination logic, and clean output into whatever format or database you need. The backend background is useful here because I understand what is happening underneath the abstraction layer. I catch failure modes before they become incidents and I build things that run without someone checking on them every morning. If you have a process that still runs on copy-paste, a data source you cannot get into, or an AI integration that needs to actually work, reach out.