2026
- January 22 - How I Reduced Cloud Costs by Over $20k — 8 Learnings
- February 10 - Migrating 4 GPU Workloads from GCP to AWS in a Month
- February 25 - Designing VPN Access for a Team That Needs Strict Compliance
- March 5 - Self-Hosting Langfuse for LLM Observability on AWS
- March 20 - How I Hardened Our EKS Stack in a Month
- August 25 - Scale-to-Zero GPU Inference on EKS: $0 Idle with Karpenter and KEDA
- August 26 - Scale-to-Zero for CPU Jobs: the KEDA In-Flight Trap
- August 27 - Airflow 3 and StarRocks on EKS: What Actually Broke
- September 20 - What GPU Time-Slicing Does Not Give You
- September 22 - Self-Hosting Langfuse on Kubernetes: 7 Gotchas
- September 24 - Auditing My Own Estimate: Langfuse from $3.4K a Month to $1.0K
- September 26 - Cutting GPU Cold Starts on EKS, with Numbers
- September 29 - Isolating GPU Pools with Value-Based Taints
- October 1 - Serving a 31B 4-bit Model on Blackwell Spot GPUs
- October 4 - Why I Chose vmagent over Grafana Alloy for Edge Devices
- October 4 - Booking g6 Capacity Before the Workday
- October 5 - What to Do After Your Cloud Credits Run Out
- October 11 - Finding a Regression in NVIDIA Container Toolkit 1.20.0