Notes from Ground Control
Engineering notes on self-hosted knowledge graphs, GraphRAG, sovereign AI memory, and what it takes to run AI in regulated environments.
15 October 2026
Running Ollama in Production: Challenges and Alternatives
Learn the limitations of running ollama in production and why frameworks like vLLM are better for high concurrency and multi-user traffic.
10 October 2026
Knowledge Graph vs Vector Database: Choosing a Retrieval Backbone
Compare knowledge graph vs vector database to decide between semantic similarity and structured reasoning for your RAG system's retrieval architecture.
29 September 2026
LLM Token CDN Caching Strategy: Why CDNs Fall Short
Learn why a standard llm token cdn caching strategy fails and how semantic prefix matching at the model layer is required for effective token caching.
24 September 2026
How to Sync AI Coding Agents with Figma for Continuous Loops
Learn how to sync ai coding agents with figma using MCP to enable bidirectional data flow and replace static handoffs with continuous design-to-code loops.
22 September 2026
Unified Interface for Multiple AI Agents: The Case for Dashboards
Learn how a unified interface for multiple AI agents reduces friction by consolidating specialized agents into a single control plane for coordination.
15 September 2026
Rootless Docker UI Tools List: Top Self-Hosted Options
Find a rootless docker ui tools list featuring Lazydocker, Portainer CE, Dockge, and Komodo. Learn which UIs support user-specific socket path configuration.
10 September 2026
How to Extract TIF Raster Coordinates in R When rasterToPoints Fa
Learn how to extract tif raster coordinates r using terra::extract() to avoid memory crashes and handle large TIF files more efficiently than rasterToPoints.
8 September 2026
Enterprise Software Moving Away from SaaS: The Shift Explained
Explore why enterprise software moving away from saas is driven by AI agents, seat-based pricing failures, and a shift toward private cloud deployments.
3 September 2026
How to Fill Time Gaps Irregular Timestamps Pandas Guide
Learn how to fill time gaps irregular timestamps pandas workflows using resample, reindex, and interpolation to standardize your time series data.
25 August 2026
How to Improve Basic FAISS RAG Pipeline Performance
Learn how to improve basic faiss rag pipeline results using recursive chunking, retrieval gates, and RAGAS evaluations to stop hallucinations.
20 August 2026
Best Tools for LLM Product Analytics: Evals and Monitoring
Discover the best tools for llm product analytics. Learn how to use Confident AI, DeepEval, and Langfuse for pre-production evals and production monitoring.
18 August 2026
Entity Resolution Duplicate Names: Improving RAG Accuracy
Learn how entity resolution duplicate names prevents retrieval fragmentation and hallucinations in RAG systems by merging records into a golden record.
9 August 2026
How to Build Knowledge Graph from Documents for Your Company
Learn how to build knowledge graph from documents using a four-stage pipeline to improve multi-hop reasoning and entity relationship extraction.
5 August 2026
EU AI Act for RAG Deployments: What Actually Applies Since August 2026
Article 50 transparency is enforceable since 2 August 2026, while high-risk duties moved to 2027. A practical checklist for teams running RAG and AI memory in the EU.
3 August 2026
The Real Cost of Metered AI Memory (and Why We Refuse to Count Tokens)
Cloud memory platforms charge around $2.50 per million tokens after a small free grant. The arithmetic looks harmless - until you run a real team's workload through it.
1 August 2026
Self-Hosted RAG in 2026: Why Serious Teams Are Moving On-Prem
Data sovereignty, predictable cost, and EU compliance are pushing RAG workloads back on-prem. What self-hosting actually takes, and what you get for it.
30 July 2026
GraphRAG vs. Vector RAG: When a Knowledge Graph Beats Top-k
Vector search answers "what sounds similar" - a knowledge graph answers "what is actually connected". Where plain RAG breaks, and when GraphRAG is worth it.