Search as a constant experimentation cycle
Vector Podcast episode with Doug Turnbull, Staff Relevance Engineer, Shopify
New episode of Vector Podcast, Season 2!
I’ve had an immense pleasure to talk to a mogul in Search world, original creator of Quepid (Search Quality Assurance and Toolbox I recommend to all my clients), Splainer, Learning to Rank plugin for Elasticsearch and an educational GitHub repo hello-ltr, author of books (Relevant Search and AI Powered Search), frequent speaker and blog writer — Doug Turnbull, Staff Relevance Engineer with Shopify.
Want to learn from Doug? Subscribe to his ML Powered Search course! I’m confident you will learn a lot through practice and building the mindset for approaching modern Search projects.
Doug’s entering search was purely accidental, like for many of us. Having been a C/C++ developer originally working on performance optimization, he found himself in Charlottesville, bumping into Eric Pugh of Open Source Connections on a developer meetup.
Doug is a very curious Search professional, heavily pushing his code in the open source and making thought contributions around Search as a field. We discussed how Quepid came about, supporting the repeating pattern of test-driven relevancy improvements. He thinks a lot about how to best design search R&D projects: to be successful, you need to be able to combine both data science and engineering sides.
We discussed (of course!) the boiling topics of dense search, sparse search, hybrid search, and ML powered search. You got it already, I’m sure: if you want to get inspired in Search and ML, listen to this episode.
Here is the list of topics we covered:
01:30 Doug’s story in Search
04:55 How Quepid came about
10:57 Relevance as product at Shopify: challenge, process, tools, evaluation
15:36 Search abandonment in Ecommerce
21:30 Rigor in A/B testing
23:53 Turn user intent and content meaning into tokens, not words into tokens
32:11 Use case for vector search in Maps. What about search in other domains?
38:05 Expanding on dense approaches
40:52 Sparse, dense, hybrid anyone?
48:18 Role of HNSW, scalability and new vector databases vs Elasticsearch / Solr dense search
52:12 Doug’s advice to vector database makers
58:19 Learning to Rank: how to start, how to collect data with active learning, what are the ML methods and a mindset
1:12:10 Blending search and recommendation
1:16:08 Search engineer role and key ingredients of managing search projects today
1:20:34 What does a Product Manager do on a Search team?
1:26:50 The magical question of WHY
1:29:08 Doug’s announcements
Also check out the show notes, where you find more details of what we covered and mentioned in the episode.
As usual, beside YouTube, you can listen on:
Spotify:
Apple Podcasts:
Or you own podcast app with the RSS feed:
https://rss.com/podcasts/vector-podcast/638830/
Remember to subscribe to stay tuned!
