MAMundher Al-Shabi, PhDinmundher.com·Aug 15 · 3 min readWhy Dynamic Reasoning Beats Static GraphsStandard vector RAG is great for finding needles, but it’s terrible at describing the haystack. It relies entirely on semantic proximity. That means it often fails at multi-hop reasoning or answering 00
MAMundher Al-Shabi, PhDinmundher.com·Jul 19 · 3 min readHow to Build Privacy into LLM Agents Without Breaking Their BrainsI've been spending a lot of time lately building autonomous agents with ReAct and RAG frameworks. The shift from stateless chatbots to stateful systems that can actually execute tools is incredibly us00
MAMundher Al-Shabi, PhDinmundher.com·Jul 4 · 3 min readStacking Agent Memory: Checkpoints, Status Boards, and Active ContextI love the idea of autonomous coding agents. But one of the quickest ways to hit a wall when setting them up is the infinite context problem. Your context window is finite, but the work you want the a00
MAMundher Al-Shabi, PhDinmundher.com·Jun 26 · 4 min readSmart LLM RoutingBuilding LLM apps is easy, but scaling them without setting a pile of money on fire is hard. You really don't need the massive brainpower of GPT-5 for every single user query. Routing is how we fix th00
MAMundher Al-Shabi, PhDinmundher.com·Jun 19 · 4 min readUsing Simulators to Evaluate Multi-Turn AI AgentsBuilding a multi-turn conversational AI is surprisingly easy right now. Evaluating it is incredibly hard. For single-turn tasks, a standard static dataset works fine: you just feed in a prompt and ass00
