Blog
Thoughts on AI engineering, DevOps, and building systems that work. No recycled advice. Just what I learn shipping things.
Most 'AI-powered' DevOps tools just send you Slack alerts. Sanskriti Labs is different — an AI that actually executes: spins up containers, optimizes queries, and scales resources while I sleep.
Researchers spend 60% of their time reading papers that don't matter. Here's how I built an AI research assistant that finds the signal in the noise — and actually surfaces new research directions.
Everything in one place — from spinning up n8n behind a Cloudflare Tunnel to sending your first WhatsApp AI reply. No cloud bills, no port forwarding, no gaps.
Install n8n in Docker, expose it via Cloudflare Tunnel, and run it at n8n.yourdomain.com without opening any firewall ports.
A Bash wrapper script to start, stop, restart, and monitor both n8n and its Cloudflare Tunnel as a single unit.
Clone OpenWA, run docker compose up -d, log into the dashboard, and understand the session state machine. Everything just works.
A tour of the OpenWA Docker data volume — sessions, sqlite DB, media files, .api-key — and a backup strategy that actually works.
Set up HTTP Header Auth in n8n, send your first WhatsApp message from a workflow, and configure a webhook trigger for incoming messages.
Three distinct n8n workflows built from a single webhook trigger: personal AI chat, group broadcast, and @agent mention handler. Pick your use case, import, configure, done.
Every endpoint the OpenWA REST API exposes — send text, images, stickers, templates, bulk broadcasts, typing indicators, message recall, and forwarding. With exact n8n HTTP Request node configuration.