Skip to main content
The hve-video-director wordmark assembled from six pipeline chips, above the line: Describe your product. Approve six checkpoints.

Playing with hve-video-director — v0.1.0 & v0.2.0: A Rename and a Rebase

Two releases at once: the rename that had to happen, and the rebase that deleted most of my own prose. v0.2.0 adds a reasoning layer, a numeric seam gate that verifies transitions instead of trusting them, and frame packets — plus the promo you can watch, which the tool made about itself.

August 4, 2026 · 6 min · Nebrass Lamouchi
A terminal window showing a real test run finishing: “Ran 233 tests in 21.283s” followed by OK

Playing with hve-video-director — v0.0.3: Real Terminals in Your Videos

v0.0.3 teaches hve-video-director to record real terminal sessions — your actual build, test, or deploy, captured automatically and dropped into the video. It also pins its CDN dependency with Subresource Integrity. A small, practical release.

June 10, 2026 · 4 min · Nebrass Lamouchi
Numbered pipeline stations — Design, Production, Audio & Render — each with the file it produces

Playing with hve-video-director — v0.0.2: New Engine & Tutorial Mode

hve-video-director’s rendering engine moves from Remotion to HyperFrames (HTML + GSAP, headless Chromium). Plus a third Tutorial mode, real motion-footage capture, and 10 brand design presets — and the update’s announcement video was, again, made by the tool itself.

June 5, 2026 · 5 min · Nebrass Lamouchi
The hve-video-director title card: the wordmark over the line “Every frame here was made by it.”

Playing with hve-video-director

Meet hve-video-director — a Claude Code skill that turns one command into a cinematic, branded product video. Six phases, real app captures, code-driven motion graphics, ElevenLabs voiceover, and Freesound music. The demo? Made by the tool itself.

April 28, 2026 · 5 min · Nebrass Lamouchi
My first contribution to the MLflow project

My first contribution to the MLflow project

My first PR to incorporate Hugging Face’s Evaluate BLEU metrics into MLflow project has been merged. This update aims to enhance text generation evaluations with BLEU metrics: a well-regarded metric, used to evaluate the quality of machine-generated text, particularly for translation tasks.

October 31, 2024 · 0 min · Nebrass Lamouchi
Playing with NVIDIA DeepStream SDK

Playing with NVIDIA DeepStream SDK

Introduction Recently, I got the chance to work on a project dealing with Computer Vision ! In the investigation phase, I discovered an interesting product: the NVIDIA DeepStream SDK ! Even when NVIDIA DeepStream was not adopted for our project 😅 I had a strong desire to discover this platform. I really cannot say if it’s a platform, a set of tools, a set of platforms, or even a full universe. 😊 ...

September 30, 2023 · 30 min · Nebrass Lamouchi