Available for new systems — Remote · Worldwide

FELU — AI Automation Expert & Software Developer

I build systems that remove the human bottleneck.

n8n pipelines, LLM integrations, and clean software — I take the repetitive work eating your week and turn it into something that runs itself.

SPECIALISMAI automation / software
STACKn8n · LLMs / JS · Python
CURRENTBooking Q3 ● 2 slots

02 — Selected work

Proof, not promises.

03 / FEATURED
/01

EammaGloberMovers — inbox → WhatsApp

● live systemn8nGmail APIWhatsApp API
$ trigger: email.received → parse → whatsapp.send … 0 missed alerts
/02

Support Pilot — AI ticket triage

LLMEmbeddingsPythonPostgres

Every ticket classified, routed and drafted before a human sees it. The queue stops being a queue.

/03 — Reporting pipeline

ReportFlow — 6 hrs/week → near-zero

● −94% manual timen8nLLM summariserSheets
All case studies — problem → process → result →

03 — Who builds this

I'm FELU. I got into software to delete work — then discovered most teams are drowning in tasks a pipeline could do better than a human.

For 5+ years I've built workflow systems, LLM integrations and internal tools for movers, service businesses and ops teams. My rule: if you do it twice a week, it should be a system — documented, monitored, boring in the best way.

Full story + experience →
0Manual reporting time cut — ReportFlow
0Missed alerts since EammaGlober pipeline live
0Automation systems shipped & maintained

04 — Capabilities

A stack, organised like a system.

CLICK TO EXPAND

Production pipelines, not demos. Every flow ships with retries, logging and a human override.

  • n8n orchestrationdaily driver
  • LLM integration / RAGOpenAI · local
  • Webhooks · APIs · cronevent-driven

Clean backends and interfaces around the automation — so the system is usable, not just clever.

  • JavaScript / TypeScriptNode · APIs
  • Pythonscripting · ETL
  • Postgres · SQLitestate that survives

Self-hosted first. Your data stays yours; deploys stay boring and repeatable.

  • GitHub · Pages · VPSship path
  • Dockerself-host n8n/LLMs
  • WhatsApp · Gmail · Sheets APIsglue layer

Signature — how I think

Every engagement runs like this.

01 / Input

The messy human trigger — an email, a ticket, a spreadsheet.

email.received →

02 / Process

Parse, classify, decide. LLM where judgment helps, rules where it doesn't.

parse() · classify() · route()

03 / Output

The action lands — message sent, sheet updated, human freed.

→ whatsapp.send · sheet.append
live demo · watch the log
› idle — press run.

05 — Result

“FELU didn't build us a chatbot. He deleted an entire category of work — alerts, follow-ups, reports just happen now.”
— FIMIOJERINDE, Operations Lead, EammaGloberMovers

06 — Next step

Have a system worth automating? Let's talk.

One call. You describe the bottleneck, I tell you honestly whether it's automatable — and what it would take.