https://banercorp.com/blog/ Notes on building production systems: AI agents, system design, cloud architecture, and data engineering — the decisions, the failure modes, and the fixes. en https://banercorp.com/blog/docling-field-test/ https://banercorp.com/blog/docling-field-test/ Fri, 30 Oct 2026 00:00:00 +0000 IBM's Docling (69,878 GitHub stars) converts complex, unstructured documents into structured data for GenAI and RAG workflows. This hands-on test runs it on a local machine against a plain-extraction baseline — Docling's own technical report, the Transformer paper, a scanned page, a speech clip, and a video — with timings, output evidence, and a Go / No-Go / Monitor verdict. https://banercorp.com/blog/beyond-the-chatbot-multi-agent-system/ https://banercorp.com/blog/beyond-the-chatbot-multi-agent-system/ Fri, 16 Oct 2026 00:00:00 +0000 I skipped the single "super-agent" and built five domain-isolated specialists on Vertex AI Agent Engine — then worked through the production hurdles that only show up at scale: orphaned sessions, silent memory failures, token bloat, and learning from transcripts without leaking PII. https://www.linkedin.com/pulse/timeline-modern-ai-decades-making-abhishek-banerjee-ew76e https://www.linkedin.com/pulse/timeline-modern-ai-decades-making-abhishek-banerjee-ew76e Sat, 11 Apr 2026 00:00:00 +0000 The real story of how modern AI came to exist: decades of ignored research, one competition, a few bold defections, and one chatbot that changed everything. 1986–2026. https://medium.com/@abhishek.banerjee219/ai-agents-to-make-sense-of-your-data-lineage-bb3863a16543 https://medium.com/@abhishek.banerjee219/ai-agents-to-make-sense-of-your-data-lineage-bb3863a16543 Sun, 22 Feb 2026 00:00:00 +0000 How I used the Model Context Protocol to build a ‘Code Analyst’ agent that traces SQL lineage across GitHub repos — and answers questions about the codebase in plain English.