athreesh

Two-Layer Knowledge Management for Claude Code

A system for persistent, cross-session memory using a git-backed work vault (raw findings) and an Obsidian vault (polished synthesis). Claude Code operates on both layers directly.

Architecture

┌─────────────────────────────────────────────────────┐
│                   Claude Code                        │
│ │

athreesh

CLU: Dynamo vs llm-d — Inference Orchestration Platform

Customer: AI-natives and CSP/NCPs deploying LLM inference at scale on Kubernetes Decision: Which orchestration layer to build on for production serving? Date: February 2026 | Dynamo 0.8.1 vs llm-d 0.5.0

---

PRODUCT CONFIGURATION

athreesh

Grove on Kind/MicroK8s: kubectl-grove CLI Demo Guide

This guide walks you through deploying Grove on a local Kubernetes cluster (Kind or MicroK8s) and testing all kubectl-grove CLI commands.

Prerequisites

• Docker installed and running
• Go 1.22+ (for building kubectl-grove)
• kubectl configured
• Kind or MicroK8s installed

athreesh

Dynamo Documentation Audit Report

Date: January 12, 2026
PR: #5380
Branch: docs/documentation-audit-updates

---

Transcript

athreesh

Claude Code PM Skills for AI Inference

Setup

gh gist clone 74fdca4ce6cf6e48711fc9b06e95dc74 ~/.claude/commands

Skills

• /explain - Teaches inference concepts in PM-friendly terms
• /architecture - Generates codebase architecture diagrams

athreesh

Dynamo: Disaggregated Serving & KV-Aware Routing Study Guide

A comprehensive guide covering disaggregated LLM inference, KV-aware routing, and benchmarking insights from the Qwen3-32B benchmark on H200 GPUs.

---

Table of Contents

1. Overview
2. Benchmark Setup
3. Results Comparison