{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Database School","title":"Building search for AI systems with Chroma CTO Hammad Bashir","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/e96e3875\"></iframe>","width":"100%","height":180,"duration":4003,"description":"Hammad Bashir, CTO of Chroma, joins the show to break down how modern vector search systems are actually built from local, embedded databases to massively distributed, object-storage-backed architectures. We dig into Chroma’s shared local-to-cloud API, log-structured storage on object stores, hybrid search, and why retrieval-augmented generation (RAG) isn’t going anywhere.\nFollow Hammad:\nTwitter/X:  https://twitter.com/HammadTime\nLinkedIn: https://www.linkedin.com/in/hbashir\nChroma: https://trychroma.com\nFollow Aaron:\nTwitter/X:  https://twitter.com/aarondfrancis \nDatabase School: https://databaseschool.com\nDatabase School YouTube Channel: https://www.youtube.com/@UCT3XN4RtcFhmrWl8tf_o49g  (Subscribe today)\nLinkedIn: https://www.linkedin.com/in/aarondfrancis\nWebsite: https://aaronfrancis.com - find articles, podcasts, courses, and more.\nChapters:\n00:00 – Introduction From high-school ASICs to CTO of Chroma\n01:04 – Hammad’s background and why vector search stuck\n03:01 – Why Chroma has one API for local and distributed systems\n05:37 – Local experimentation vs production AI workflows\n08:03 – What “unprincipled data” means in machine learning\n10:31 – From computer vision to retrieval for LLMs\n13:00 – Exploratory data analysis and why looking at data still matters\n16:38 – Promoting data from local to Chroma Cloud\n19:26 – Why Chroma is built on object storage\n20:27 – Write-ahead logs, batching, and durability\n26:56 – Compaction, inverted indexes, and storage layout\n29:26 – Strong consistency and reading from the log\n34:12 – How queries are routed and executed\n37:00 – Hybrid search: vectors, full-text, and metadata\n41:03 – Chunking, embeddings, and retrieval boundaries\n43:22 – Agentic search and letting models drive retrieval\n45:01 – Is RAG dead? A grounded explanation\n48:24 – Why context windows don’t replace search\n56:20 – Context rot and why retrieval reduces confusion\n01:00:19 – Faster models and the future of search stacks\n01:02:25 – Who Chroma is for and when it’s...","thumbnail_url":"https://img.transistorcdn.com/QPpKhRBWtM56k1sC7Y7j7aqGy8sd-XKWw6mEKgA1jmw/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9lMmRl/ZTM5ZTAxOTMzNTgy/MmJmZTA4MWE1ZjMw/YjE4Ny5wbmc.webp","thumbnail_width":300,"thumbnail_height":300}