M K Pavan Kumar·Oct 30Does Index-Native BM25 Change Retrieval Quality? A Two-Collection Experiment in Qdrant 1.20What if you could remove the separate BM25 sparse-encoding step from your hybrid RAG pipeline? Would you still retrieve the evidence that…
InDevOps.devbyM K Pavan Kumar·Oct 17Evaluating Medical RAG with Qdrant, Jev and MLflow — A Production Ready SetupWe often focus on retrieval accuracy and response quality, but what happens when we need to evaluate every answer against ground truth…
M K Pavan Kumar·Oct 5Catching Hallucinations Before They Ship: A Claim-Level Verdict Gate for Clinical RAGRAG answers over clinical guidelines can sound confident while citing the wrong passage or drifting from what the source says, and in a…
InTowards DevbyM K Pavan Kumar·Sep 26Slicing and Prefixing a Medical Knowledge Base: Two Small Qdrant Features That Solve Big…A few months ago I loaded the MedQuAD corpus into Qdrant to build a patient-education assistant. Nothing fancy: 7,936 medical explainer…
In𝐀𝐈 𝐦𝐨𝐧𝐤𝐬.𝐢𝐨byM K Pavan Kumar·Sep 24Creating coding agents with AgnoMost coding assistants stop after generating code. But real software engineering does not work that way. Code needs to be written…
InGoPenAIbyM K Pavan Kumar·Sep 6RAG Beyond Retrieval: Model Registry, Vector Search and Intelligent LLM RoutingThese days, most RAG applications are built around a single embedding model, a single knowledge base, and a single LLM. But not every model…A response icon1A response icon1
InDevOps.devbyM K Pavan Kumar·Aug 20MLflow MCP Registry in Action: An End-to-End Hands-On GuideIn this article, we will see how to take the MLflow MCP Registry from concept to a complete working setup. We will build MCP servers, test…
InStackademicbyM K Pavan Kumar·Aug 16Observable Voice RAG Agents: Tracing RAG Agents End-to-End with MLflow and QdrantToday in this article, we shall look at how to build a voice-enabled RAG agent that can interact with users in real time, retrieve relevant…A response icon1A response icon1
InTowards DevbyM K Pavan Kumar·Aug 6Dense + Sparse + Filter: Building Precision Retrieval for Annual Reports and 10-KsToday we are going to see how a simple idea, teaching your retrieval system to respect boundaries instead of just chasing meaning, can fix…
InStackademicbyM K Pavan Kumar·Jul 22Hybrid Retrieval Under the Microscope: BM25 vs MiniCOIL on MedQuADToday, we are going to learn how BM25 and miniCOIL perform in a hybrid search pipeline. Using the MedQuAD dataset, EmbeddingGemma, and…
