A data-driven analysis of Cole Medins video on AI replacing software engineers with added perspective from daily AI coding tool usage
Cleaning up the legacy skills directory, adding commands/ for slash command autocomplete, fixing per-language cover image generation, and bumping to v0.2.5
Replacing VEO 3 with DashScope Wan 2.2 for cheaper video generation, and building SAM 2.1 interactive background removal to replace rembg
How to run SAM 2.1 locally on Apple Silicon Mac with MPS GPU acceleration and a comparison with Meta's latest SAM 3 online playground
Adding ATR-based dynamic stop-loss, investment horizon parameter, and position re-evaluation logic while fixing investor inquiry parameters and portfolio double-counting bugs
A comprehensive look at the AI video and image generation API market in early 2026 covering pricing comparisons across fal.ai, Replicate, OpenAI, Google Veo 3, Wan 2.1, and GPT Image 2 with practical cost optimization strategies
A deep dive into Claude Code ultra plan mode, Karpathy's Obsidian-based RAG alternative, self-evolving agent memory, and 12 rules for context optimization
First dev log of a Korean-language multi-agent system for advertising creative work — a static mockup of the four-stage chat-first pipeline (research, copy, scenario, storyboard) and the design language it would carry into the React rewrite six weeks later
Building a Grafana resource dashboard with OpenTelemetry metrics, then fixing CPU/RAM spikes discovered through trace analysis using thread executors and timeouts
Building a post-processing pipeline to remove backgrounds from VEO-generated animation frames using rembg, binary alpha thresholding, color decontamination, and edge refinement to produce transparent APNGs
