AI Automation Engineer · Open Source

Always-on AI systems that work while you sleep

I turn manual business processes into production AI automation: 24/7 agents, Telegram and WhatsApp bots, and data pipelines that run on free-tier cloud. Everything here is built from real incidents on a live box, not from tutorials.

Try AI Can Run → Self-Hosted Agent Stack GitHub profile
24/7Agent uptime on free tier
$0–5Monthly infra cost
2 OCPU · 12 GBProduction envelope
100%Client-side detection

Selected work

AI Can Run

Browser-side hardware detection that tells you which open models your own machine can run — chat, coding, image and video, ranked by real VRAM math.

Live tool · no install

Self-Hosted Agent Stack

The operational playbook for a 24/7 agent on a free ARM box: watchdogs, zram tuning, stray-process reaping, crash recovery and the lessons that cost a real outage.

Infrastructure · bash + python

LinkedIn Autopilot

AI drafts your posts, you approve, it publishes — approvals stay in the loop, content stays local-first and free of SaaS fees.

Content automation

Multi-Agent Research Pipeline

Crashproof multi-agent orchestration: researchers, reviewers and writers with resumable state, so a failure at step 9 does not restart the job.

Agent engineering

Hermes Slack Agents

One gateway, per-client agents: every client gets an isolated bot and isolated memory, self-hosted, no per-seat fees.

Multi-tenant ops

Voice Portfolio

A talking portfolio: the page answers out loud. On-device browser voice agent in English and Arabic editions, with a voice CV, a video CV and a downloadable CV — no backend.

Live demo · voice AI

How I work

  1. Measure first — RSS, disk, latency, cost. No optimisation without a before number.
  2. Stay inside the free tier, because the free tier is the whole advantage.
  3. Make it recover itself — watchdogs, health checks, idempotent scripts, tested restores.
  4. Two languages only: the server speaks bash and Python; the client-facing layer stays simple.

Contact

Open to AI integration, automation pipelines, agent infrastructure and self-hosted deployment work. Fastest route: a DM on LinkedIn.