How Flipkart made their type ahead search hyper personalized

Build Your Own Redis / DNS / BitTorrent / SQLite - with CodeCrafters.

In the video, I discussed how Flipkart personalized its type-ahead search suggestions, focusing on high-level architecture and key design decisions. To enhance user experience, personalized suggestions based on user history were prioritized. Factors such as suggestion quality, performance, and grammar were considered for ranking. User intent was analyzed using product taxonomy, past searches, and browsing history. Personalization was achieved by training a model using user interactions. The system utilized XGBoost with LTR ranking in a search engine like Solar for efficient suggestion delivery.

# Recommended videos and playlists

If you liked this video, you will find the following videos and playlists helpful

# Things you will find amusing

# Other socials

I keep writing and sharing my practical experience and learnings every day, so if you resonate then follow along. I keep it no fluff.

Thank you for watching and supporting! it means a ton.

I am on a mission to bring out the best engineering stories from around the world and make you all fall in
love with engineering. If you resonate with this then follow along, I always keep it no-fluff.

[kamal-xdid]

Learning system design from such a experienced person is something I recommend and not from just a college passout and somehow managed to get into Google and FB and started teaching system design. As this only gets better with experience. No shortcut!

[parthpathak]

I have just recently work on hibernate search apache lucene for indexing & searching the query results.
In this, we used fuzzy & wildcard query to search in an lucene index.

Your videos help us to clear the basics of any tech stack.👍

[sagar-ttub]

Great video Arpit. Your knowledge never ceases to leave me in awe

[deepakoraon]

Never thought typeahead were this advanced

[sauravsingh]

my understanding of type ahead was so limited, thanks a lot arpit.

[beopinioned]

Arpit, you have done a very good job. Thank you.

[newbie]

17:35 ah, this is what the Query Understanding services do
Mis-spell correction, Normalization, Tokenization, Category Prediction
pretty amazing stuff, would love to have an in-depth analysis of the ML side of these things as well

[hawkeyeyt]

Hey Arpit, thanks for making awesome content. I don't know why but I find it hard to focus when I see the content written in this type of font. Nevertheless, keep making these videos.

[noads]

Netflix does this performance thingy...where it show's movies in search which are not available....does this mean, they're simply using some public search data source for ES

[akhileshit]

Thanks a ton for this Arpit, would like to know more about how taxonomy /RDF being used in real search systems.
Also, apart from Solr in LTR mode, do we really need graph DB like neo4j for enabling semantic search. and answering complex search

[haridotvenkat]

Any idea how they provide the results for an user that didn't login? Cache layer is one obvious answer. location may be another parameter for suggestion. IP address?!!!

[sounishnath]

Can you please share the Reformulating search Model building tutorial. That’s fascinating 😮