Notes from production.
Technical deep dives on search relevance, retrieval, RAG and LLM training, written by founder Monadhika Sharma from experience building these systems at scale.
Hybrid search: how to combine BM25 and vector search without losing the best result
Hybrid search done right: why merging BM25 and vector results can bury your best match, what true hybrid scoring looks like, and benchmark results.
Why embeddings don't work for your search: 7 failure modes and how to fix them
When embeddings don't work, they fail in predictable ways: negation, numbers, IDs, hard constraints and domain jargon. How to spot each and fix it.
Free 30-minute intro call
Have a search or LLM problem worth solving?
Tell me what you're building. You'll talk with me directly, not a sales team. If I'm not the right fit, I'll say so and point you to someone who is.
or email hello@infersense-labs.ai