node01 just went from one NVIDIA RTX PRO 6000 Blackwell to two. The motivation was simple: a single 96GB card is plenty for shared inference and small-to-medium training jobs, but it can't run true multi-GPU parallel training. With model sizes creeping up, we needed a second card so people can actually spread a training job across GPUs instead of being capped at one.

After replacing the screen on a Google Pixel 8, the under-display fingerprint sensor needs to be recalibrated. Google provides an official tool at pixelrepair.withgoogle.com/udfps, but it only works on Windows and macOS — and on newer firmware versions it may report "firmware too new" and refuse to run.

This guide documents a working method using the command line on Linux, bypassing the browser tool entirely.

I recently installed two Samsung 970 EVO Plus 1TB NVMe drives into our QNAP TS-855X NAS as a read/write cache layer. The result: write speed jumped from 169 MB/s to 673 MB/s — nearly 4×.

After my previous experiments with camera-based detection and FLOAM-based LiDAR odometry in CARLA, I wanted to push things further — this time focusing on robustness under adversarial conditions.

After my previous CARLA experiment with camera-based object detection, I wanted to push further — this time focusing on LiDAR and localization.

I recently started experimenting with the CARLA simulator and decided to test whether a pre-trained perception model — with zero exposure to simulated data — could hold up against synthetic imagery.

After completing my SCNN lane detection project, I reviewed the RESA paper (2020), which extends SCNN's message-passing mechanism with a Recurrent Feature Shift Aggregator (RESA) and a Bilateral Up-Sampling Decoder (BUSD). The paper claimed a 9% F1-score improvement over SCNN. My experiments told a different story.

I recently worked on a lane marking detection project. After reviewing the literature, I decided to use the SCNN (Spatial CNN) model — not because it's state of the art, but because it falls squarely into my "if you can't implement it, you don't really understand it" category.

Over the past few weeks, I've been working on a Python tool that converts Synthesia-style piano videos into MIDI files using computer vision — and I'm making it public today.

I recently deployed three deep learning models covering three core perception tasks — all running simultaneously in real-time on a single GPU.