Ketan Bhatt

Engineering @intercom

[Ketan Bhatt @_ketanbhatt]

In hindsight, 2019 was an exceptionally eventful year for me. Traveled to 3 countries, & 3 states in India; took 4 months off; shifted to Dublin, living on my own for the first time; for @intercom, my first ever largeish company. Extremely grateful for this. Thank you, fate(?)!

[Fergal Reid @fergal_reid]

Really excited that Brett Chen from @perplexity_ai will be joining us as a speaker at our AI event next week.
He'll talking about about "Scaling Intelligence: Production-Grade AI Agents and Models at Perplexity".
Live in SF, and streaming.
Details in thread.

[Ketan Bhatt @_ketanbhatt]

@arindambarman @Tarun_Garg2 might be able to help.

[Ketan Bhatt @_ketanbhatt]

@SagarJoglekar And good morning to you too brother, weekend is right around the corner.

[Ketan Bhatt @_ketanbhatt]

@vortex_ape Uv is amazing

[Ketan Bhatt @_ketanbhatt]

@xeraa @guidsen @elastic @IntercomEng @brian_scanlan @fergal_reid Yes, with 80k docs, and 200ish per tenant, brute force will be blazingly fast.

Sentences might have too little information to give any benefit when encoded. I recommend starting out with paragraphs, maybe break them into smaller chunks if paragraphs are too long. And iterate.

[Ketan Bhatt @_ketanbhatt]

@guidsen @xeraa @elastic @IntercomEng @brian_scanlan @fergal_reid You can choose to do these steps yourself, or just let Elasticsearch do it (as Philipp mentioned).

[Ketan Bhatt @_ketanbhatt]

@guidsen @xeraa @elastic @IntercomEng @brian_scanlan @fergal_reid like just chunking by paragraph. If you have less than 100k documents, I would suggest you don't even need to bother with using an ANN search: just use script_score and run an exact KNN search.

[Ketan Bhatt @_ketanbhatt]

I wrote about our experience using Elasticsearch for running Fin.

[Fergal Reid @fergal_reid]

A new entry on our Intercom AI blog from @_ketanbhatt, about our decision to stick with @elastic as opposed to adopting a new VectorDB for Fin.
tl;dr: operational knowledge is valuable, and ES will do vectors.
Link below!

[Ketan Bhatt @_ketanbhatt]

@0xferruccio @vortex_ape @adisbanda Looks amazing folks! Well done 🍻

[Jack Ryan @jmfryan]

I’ve been working on ways to integrate our AI agent Fin into pretty much anything.

It means folks can use the full power of Fin, including workflows, actions, etc, wired up to anything they like - the command line, a cmd+K interface, their own messenger.

Beta coming soon!

[Ketan Bhatt @_ketanbhatt]

@mariokostelac @pratik60 We too Mario, we too 😂

[ISRO @isro]

Hat-trick for ISRO in RLV LEX! 🚀

🇮🇳ISRO achieved its third and final consecutive success in the Reusable Launch Vehicle (RLV) Landing EXperiment (LEX) on June 23, 2024.

"Pushpak" executed a precise horizontal landing, showcasing advanced autonomous capabilities under challenging conditions.

With the objectives of RLV LEX accomplished, ISRO embarks into RLV-ORV, the orbital reusable vehicle.

@IAF_MCC @DRDO_India @CSIRNALOFFICIAL
@IITKanpur @IndianOilcl @AAI_Official

[Ketan Bhatt @_ketanbhatt]

@adrian_cooney Looks class!

[Ketan Bhatt @_ketanbhatt]

@mariokostelac 😂

[Mario Kostelac @mariokostelac]

TIL To figure out the dependency tree for a specific package, https://t.co/aoeFepgCUt is quite faster than reading pip-compiled requirements.txt.

[Ketan Bhatt @_ketanbhatt]

@arindambarman @zmiro Looks so good 👏🏼

[Ketan Bhatt @_ketanbhatt]

@Tarun_Garg2 Are these people you know? If not, it might just be the Twitter algorithm showing you tweets similar to tweets of this person who wrote a sabbatical handbook 😀

[Fin @fin_ai]

Time to have 'THE CX TALK' with your support bot 😘

[Ketan Bhatt @_ketanbhatt]

@shilpiagrawal55 Probably because they both have an ensemble and are magical...