> initializing ast_labs...

> loading highlight-desk...

> connecting n8n webhooks...

> system ready.

Independent engineer. Connected systems.

Automating
the impossible.

AST Labs is a technical studio based in Bartlett, TN. I engineer custom macOS media pipelines, Python web scrapers, and autonomous AI lead capture systems.

Initialize Project
↳

Less manual.
More possible.

Inspect the systems
Built with the right tools.
[ Python ][ FFmpeg ][ Whisper AI ][ Ollama ][ Playwright ][ Make.com ][ n8n ][ Twilio 10DLC ][ Vapi.ai ][ Tailwind CSS ]
~/ systems

Inside the command center.

Local engines. Connected workflows.
Built around the work that needs doing.

macOS / media engineering

Highlight Desk

A macOS recorder that saves the moments Twitch chat reacts to, reframed as vertical clips and filed locally, in Dropbox or in Google Drive.

Node.jsStreamlinkFFmpeg
Runs on one Mac. Retries failed uploads.
Highlight Desk macOS app showing live capture controls and recent uploaded clips
01 / Live capture Actual macOS app
Highlight Desk clip library with chat-triggered highlights and successful Dropbox upload statuses
02 / Clip library Dropbox upload history
Bartlett Website Design

Universal Missed-Call Text-Back

A 38-node n8n workflow for contractors and auto shops. Conversation rules are plain code, checked by 21 offline tests.

Workflow preview
Universal_Missed_Call_Text_Back.jsonCanvas
TriggerMissed CallBusy or no answer
ReceiveTwilio WebhookHTTP / POST, authenticated
ProcessIntake rules by trade
& SMS outbox
Ask → record → hand off to staff
JavaScript / Media Tools

Highlight Desk

Streamlink pulls the stream and FFmpeg copies it into 3-second segments. Chat is sampled every 5 seconds; when it spikes, the recorder saves 15 seconds before and 15 after, then reframes the clip to 1080×1920 with the full frame centered over a blurred fill. Clips stay on the Mac or upload to Dropbox or Google Drive.

Node.jsStreamlinkFFmpegDropboxGoogle Drive
  1. 01Record
  2. 02Watch chat
  3. 03Reframe
  4. 04Store

The recorder reconnects after a stream drop, retries failed uploads across restarts and writes a JSON record beside every clip. It ships as a macOS app with its own FFmpeg and Streamlink runtime.

The actual FFmpeg filter builder: split the frame, blur the background and center the foreground.
function buildReframeFilter({ width, height, blur }) {
  const halfWidth = Math.max(2, Math.floor(width / 4) * 2);
  const halfHeight = Math.max(2, Math.floor(height / 4) * 2);
  return [
    "[0:v]split=2[bg][fg]",
    `[bg]scale=${halfWidth}:${halfHeight}:force_original_aspect_ratio=increase,crop=${halfWidth}:${halfHeight},gblur=sigma=${blur}:steps=5,scale=${width}:${height},setsar=1[bgout]`,
    `[fg]scale=${width}:${height}:force_original_aspect_ratio=decrease,setsar=1[fgout]`,
    "[bgout][fgout]overlay=(W-w)/2:(H-h)/2:format=auto,format=yuv420p[v]"
  ].join(";");
}
~/ next_project

Your bottleneck.
My next build.

Tell me what you’re doing manually.
Let’s engineer a better way.

Bartlett, Tennessee
Built locally. Connected everywhere.
POST /api/initiate-contact

Your details are used to respond to this project inquiry.