<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Home</title><description>Distributed systems and infrastructure engineer building AI-powered systems.</description><link>https://mister-raggs.github.io/</link><item><title>One scene, three futures: the GPU kept up, but the choices did not always</title><link>https://mister-raggs.github.io/blog/one-scene-three-futures/</link><guid isPermaLink="true">https://mister-raggs.github.io/blog/one-scene-three-futures/</guid><description>I batched three five-second video futures from one frame on a B200, built a choice-and-continue interface, and found that fast generation and controllable outcomes are very different things.</description><pubDate>Sun, 27 Sep 2026 00:00:00 GMT</pubDate></item><item><title>One Scene, Three Futures</title><link>https://mister-raggs.github.io/projects/three-futures-research/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/three-futures-research/</guid><description>Branching video generation on one B200 — three five-second futures from a single frame in 7.37 s, a choose-and-continue loop, and a measured line between generating fast and generating what was asked.</description><pubDate>Sun, 27 Sep 2026 00:00:00 GMT</pubDate></item><item><title>231s to 192s: 4-bit video diffusion on a $3,000 desktop</title><link>https://mister-raggs.github.io/blog/fp4-video-diffusion-dgx-spark/</link><guid isPermaLink="true">https://mister-raggs.github.io/blog/fp4-video-diffusion-dgx-spark/</guid><description>Quantizing a video model&apos;s linear layers to 4-bit cut a 1080p generation by 39 seconds — and did nothing at all on the next two models I tried it on. The gap is the interesting part.</description><pubDate>Sun, 23 Aug 2026 00:00:00 GMT</pubDate></item><item><title>OneBusAway / maglev: OSS Contributions</title><link>https://mister-raggs.github.io/projects/onebusaway-maglev/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/onebusaway-maglev/</guid><description>Upstream contributions to OneBusAway&apos;s Go REST API — mutex contention profiling under k6 load, real-time detector fixes, and a versioned San Diego GTFS ground-truth store.</description><pubDate>Tue, 14 Jul 2026 00:00:00 GMT</pubDate></item><item><title>From 15 minutes to 45 seconds: rebuilding an Amazon on-call tool in Go</title><link>https://mister-raggs.github.io/blog/building-log-triage/</link><guid isPermaLink="true">https://mister-raggs.github.io/blog/building-log-triage/</guid><description>How I rebuilt a production log triage system — binary search over a sorted index, a channel-based ingestion pool, a write-ahead log for crash recovery, and what I&apos;d do differently at scale.</description><pubDate>Wed, 22 Apr 2026 00:00:00 GMT</pubDate></item><item><title>JobHunter</title><link>https://mister-raggs.github.io/projects/jobhunter/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/jobhunter/</guid><description>Automated job scraper monitoring 40+ companies across 8 ATS platforms every 30 minutes, deduplicates postings in SQLite, and emails new listings via Resend API — deployed on a DigitalOcean VPS with a live dashboard.</description><pubDate>Thu, 09 Apr 2026 00:00:00 GMT</pubDate></item><item><title>Log Triage v2: Sub-Millisecond Incident Log Search</title><link>https://mister-raggs.github.io/projects/log-triage-v2/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/log-triage-v2/</guid><description>Go-based log triage system that finds the closest log entry to an incident timestamp in sub-millisecond time across millions of log lines — reimagined from a production system at Amazon.</description><pubDate>Fri, 20 Mar 2026 00:00:00 GMT</pubDate></item><item><title>Building Flare: LLM-Powered Incident Detection on Real Log Data</title><link>https://mister-raggs.github.io/blog/building-flare/</link><guid isPermaLink="true">https://mister-raggs.github.io/blog/building-flare/</guid><description>How I built an end-to-end log anomaly detection pipeline that combines classical ML with LLM summarization — and what I learned about evaluating LLM output in production-adjacent systems.</description><pubDate>Tue, 17 Mar 2026 00:00:00 GMT</pubDate></item><item><title>Flare: LLM-Powered Log Anomaly Detection</title><link>https://mister-raggs.github.io/projects/flare/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/flare/</guid><description>End-to-end log anomaly detection pipeline with multi-model comparison (Isolation Forest, LOF, One-Class SVM), MLflow experiment tracking, and LLM summarization on real HDFS log data.</description><pubDate>Tue, 17 Mar 2026 00:00:00 GMT</pubDate></item><item><title>Parade: VC-Funded Job Discovery Pipeline</title><link>https://mister-raggs.github.io/projects/parade/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/parade/</guid><description>Automated job discovery pipeline that monitors VC funding news, resolves companies via Clearbit with graceful degradation, scrapes careers pages across Lever/Greenhouse/Ashby, and delivers a daily digest of tech roles by email.</description><pubDate>Sun, 01 Mar 2026 00:00:00 GMT</pubDate></item><item><title>Meeting-to-PR: Voice-Driven Code Generation</title><link>https://mister-raggs.github.io/projects/meeting-to-pr/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/meeting-to-pr/</guid><description>Voice-driven code generation system that transforms verbal developer conversations into production-ready GitHub pull requests via an AI-powered Discord bot.</description><pubDate>Sat, 24 Jan 2026 00:00:00 GMT</pubDate></item><item><title>GPU Rental AI Agent</title><link>https://mister-raggs.github.io/projects/agentic-gpu-renter/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/agentic-gpu-renter/</guid><description>AI agent that automatically finds, rents, and manages cloud GPUs across multiple vendors with intelligent failure recovery and machine-to-machine payments.</description><pubDate>Sat, 10 Jan 2026 00:00:00 GMT</pubDate></item><item><title>LLM Fine-tuning for Patient Routing</title><link>https://mister-raggs.github.io/projects/llm-patient-routing/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/llm-patient-routing/</guid><description>Clinical decision support system using BioBERT fine-tuning with AWS Bedrock and FastAPI. Achieved 91.6% accuracy across 1,000+ assessments with optimized inference.</description><pubDate>Wed, 10 Sep 2025 00:00:00 GMT</pubDate></item><item><title>PyTorch RAG</title><link>https://mister-raggs.github.io/projects/pytorch-rag/</link><guid isPermaLink="true">https://mister-raggs.github.io/projects/pytorch-rag/</guid><description>Retrieval-Augmented Generation framework built with PyTorch for context-aware question answering with FAISS-based vector retrieval.</description><pubDate>Mon, 01 Sep 2025 00:00:00 GMT</pubDate></item></channel></rss>