Information Retrieval on Bharat Singhhttps://bsbhadwal.com/tags/information-retrieval/Recent content in Information Retrieval on Bharat SinghBharat Singhhttps://bsbhadwal.com/images/bharat-singh-ai-systems-architect.jpghttps://bsbhadwal.com/images/bharat-singh-ai-systems-architect.jpgHugoenTue, 13 Oct 2026 00:00:00 +0530Reciprocal Rank Fusion vs. Cross-Encoders: Designing the Sub-50ms Reranking Funnelhttps://bsbhadwal.com/posts/reciprocal-rank-fusion-vs-cross-encoders-reranking-funnel/Mon, 12 Oct 2026 00:00:00 +0530https://bsbhadwal.com/posts/reciprocal-rank-fusion-vs-cross-encoders-reranking-funnel/Designing high-throughput hybrid retrieval reranking funnels combining Reciprocal Rank Fusion (RRF) and Cross-Encoder neural scoring under sub-50ms SLAs.ColBERT and Late Interaction at Scale: Why Multi-Vector Retrieval Strains Production Memoryhttps://bsbhadwal.com/posts/colbert-late-interaction-at-scale-memory-bandwidth-physics/Wed, 22 Jul 2026 00:00:00 +0530https://bsbhadwal.com/posts/colbert-late-interaction-at-scale-memory-bandwidth-physics/Deep architectural analysis of ColBERT late interaction multi-vector retrieval, memory bandwidth physics, PLAID indexing, and sub-50ms latency SLAs.Evaluating RAG Without Ground Truth: Production Pitfalls of LLM-as-a-Judgehttps://bsbhadwal.com/posts/evaluating-rag-without-ground-truth-llm-as-a-judge-pitfalls/Mon, 15 Jun 2026 00:00:00 +0530https://bsbhadwal.com/posts/evaluating-rag-without-ground-truth-llm-as-a-judge-pitfalls/Production analysis of evaluating enterprise RAG pipelines without human ground truth, mitigating LLM-as-a-judge bias, and building empirical synthetic test sets.The Retrieval Fallacy: Why Massive Context Windows Don't Eliminate the Need for IRhttps://bsbhadwal.com/posts/the-retrieval-fallacy-why-more-tokens-dont-equal-better-context/Thu, 12 Mar 2026 00:00:00 +0530https://bsbhadwal.com/posts/the-retrieval-fallacy-why-more-tokens-dont-equal-better-context/Why 1M+ token context windows do not replace Information Retrieval. Examining needle-in-a-haystack decay, attention dilution, and cost economics.The Paradigm Shift to Late Chunking in Retrieval-Augmented Generation: A Technical and Comparative Analysishttps://bsbhadwal.com/posts/late-chunking-in-retrieval-augmented-generation/Mon, 09 Feb 2026 00:00:00 +0530https://bsbhadwal.com/posts/late-chunking-in-retrieval-augmented-generation/Technical breakdown of Late Chunking in RAG architectures. Contrasting Jina AI token mean pooling and Voyage AI contextualized embeddings against naive chunking.