AS

ABM Saifullah

Automation Specialist
Dhaka, BD — 00:00:00 GMT+6
Automation Specialist • Selected Systems & Practice

Designing systems that think, connect, and run unattended.

I am Saifullah — an Automation Specialist. I do not view automation as merely scripting tasks; I treat it as designing living architecture.

My days are spent orchestrating AI-based automated smart farming systems that read sensor physics and regulate greenhouse irrigation, engineering production n8n workflow pipelines that eliminate manual enterprise operations, and exploring the frontiers of generative AI video.

I • Core Practice

The Three Arenas of Automation

A curated exploration into physical agritech automation, enterprise workflow orchestrations, and synthetic generative media.

Agriculture has historically relied on manual labor, guesswork, and delayed human reactions. As an automation specialist, my objective is to make biological systems self-monitoring and self-regulating.

In our automated smart farming installations, microcontrollers stream soil moisture and Vapor Pressure Deficit (VPD) metrics over MQTT. Computer vision algorithms continuously audit foliage colorimetry to detect stress before it is visible to the naked human eye.

When a moisture deficit is detected, solenoid valves actuate autonomously without requiring human intervention. It is precise, water-conserving, and operates 24/7.

Smart Agriculture Rig Active Telemetry
IoT Nodes ESP32 & MicroPython edge transmitters running low-power loops.
Vision Model Computer vision foliage health & pest detection routines.
Protocol MQTT payload broker pushing telemetry to local cloud relays.
Actuation Calibrated relay circuits for threshold-governed solenoid irrigation.
Soil Moisture
42.8%
Target: 40-45%
Ambient Temp
24.6°C
Greenhouse 01
Valve State
STANDBY
Auto-Governed

Repetitive human data transfer is a symptom of fragmented architecture. When staff spend hours shuttling data between customer forms, CRMs, spreadsheets, and messaging apps, business velocity stalls.

I build self-hosted n8n workflow pipelines that operate as the silent, unattended central nervous system for businesses. These pipelines ingest webhooks, consult Large Language Models (Claude 3.5 Sonnet, GPT-4o) to evaluate context and classify intent, and synchronize records across PostgreSQL and Supabase.

The result: automated WhatsApp and Telegram client updates, zero manual data entry, and fail-safe error monitoring deployed cleanly via Docker.

Pipeline Blueprint Unattended Node Stack
Orchestrator Self-hosted n8n in containerized Docker environments.
Reasoning Claude 3.5 Sonnet & GPT-4o for entity extraction & classification.
Data Store Supabase & PostgreSQL for relational sync and vector storage.
Client Nodes WhatsApp Business Cloud API, Telegram bot dispatchers, Webhooks.

Beyond operational logic and sensors lies synthetic creative technology. Generative video is redefining cinematography, enabling single creators to conceptualize and direct cinematic motion sequences directly through code and prompt engineering.

I experiment with generative video toolchains — guiding diffusion physics with Runway Gen-3 and Kling AI, generating consistent visual keyframes with Midjourney and Flux, and designing spatial voice tracks using ElevenLabs.

It is an ongoing exploration in narrative automation: turning abstract concepts into high-fidelity audiovisual storytelling.

Generative Suite Synthetic Direction
Motion Synthesis Runway Gen-3 Alpha, Kling AI, Luma Dream Machine.
Concept Frames Midjourney v6, Flux.1 for artistic visual direction.
Neural Voice ElevenLabs speech cloning and dynamic sonic pacing.
Pipeline Automated assembly scripts and timeline formatting.
II • Principles

The Automation Creed

How I approach building systems, whether for agricultural hardware or business logic.

01 /

Unattended by Default

If an automated pipeline requires daily manual intervention to function, it is not an automated system. True automation operates quietly in the background with robust fallback mechanisms.

02 /

Physical Meets Digital

Software is most compelling when it touches the real world. Connecting real-world physical soil probes and actuators to cloud telemetry grounds high-level code in tangible physical reality.

03 /

Human Discretion

Automation should eliminate drudgery, not human judgment. The goal is to clear away mechanical repetition so humans can spend their attention on creative and strategic decisions.

III • Arsenal

Tools & Technologies

The hardware, software engines, and models I deploy in production.

n8n (Self-Hosted) Docker & Compose Claude 3.5 Sonnet OpenAI GPT-4o Google Gemini ESP32 & MicroPython MQTT Protocols OpenCV Vision Supabase PostgreSQL Runway Gen-3 Kling AI ElevenLabs WhatsApp Cloud API Linux VPS
IV • Direct Inquiries

Let's discuss automated systems.

Whether you are looking to automate an agricultural environment, design unattended n8n business pipelines, or explore creative AI video — write to me directly. No agency middlemen, no automated responses.

Copied to clipboard ✓