Vikram Oberoi

Creator @ @ Curator @

[Vikram Oberoi @voberoi]

Today is my last day at @datalabto!

I have very much enjoyed my time here. We punch so far above our weight at @datalabto that it could feel surreal. I love working on small teams and I had a lot of fun being a part of this one.

Much thanks to @VikParuchuri for the opportunity.

[Vikram Oberoi @voberoi]

@jamesdillard @citymeetingsnyc Neat! I have some exciting things to show in the next couple of weeks…

[Vikram Oberoi @voberoi]

@kaseyklimes I unfortunately remember this all winter too.

[Vikram Oberoi @voberoi]

@jamesdillard @datalabto @VikParuchuri This!

https://t.co/DsMiuAJaa1

[Vikram Oberoi @voberoi]

I’m leaving Datalab to do two things:

1. I’m going to go broad with @citymeetingsnyc and will be relaunching it under a new brand with a commercial offering this summer. It was a successful experiment and my aims are bigger and more ambitious for it today.

2…

[Vikram Oberoi @voberoi]

Today is my last day at @datalabto!

I have very much enjoyed my time here. We punch so far above our weight at @datalabto that it could feel surreal. I love working on small teams and I had a lot of fun being a part of this one.

Much thanks to @VikParuchuri for the opportunity.

[Vikram Oberoi @voberoi]

Shout out of the rest of the team on this app — @_tarunmenta, @supershishy — and those of us with the wisdom to remain off it: Vlad, Alex, and Vanessa.

[Vikram Oberoi @voberoi]

2. I’ll be available on a consulting basis again soon. While it may look different, I’ve worked with a number of startups and you can read about my work here: https://t.co/DPCsyUYWRC

[Datalab @datalabto]

Datalab is now free to start, with pay as you go 🙌

We heard your interest in a flexible way to test on smaller volumes before scaling up — no fixed monthly commitment. So we built it so you can start parsing for free.

[Vikram Oberoi @voberoi]

@_blahblahblah I love this joke.

[Datalab @datalabto]

An archivist and former engineer is digitizing 800K pages of Gold Rush history — 1850s court cases, mining claims, land deeds — and making it searchable. The challenge? Many of them are written in 19th-century cursive.

Chandra reads the handwriting his colleagues call "hieroglyphics." A 200-page case that took 3 weeks to transcribe by hand now takes 2 hours.

How the Nevada County Historical Archive did it: https://t.co/o14vBYS7f1

[Vik Paruchuri @VikParuchuri]

We're launching turbo mode data extraction - 5x faster, 5x cheaper, and 7% more accurate than Azure Content Understanding.

4.5s p50/7s p90 across 1-30 page docs - good enough for realtime user flows.

[Vikram Oberoi @voberoi]

@harryh @tqbf Whoaa impressive

[Vikram Oberoi @voberoi]

@harryh @tqbf Only managed to ID about half and had to reveal the rest -- blue and purple were hard!

[Vikram Oberoi @voberoi]

@tqbf Ok this is pretty crazy.

[Datalab @datalabto]

Releasing Chandra 2.1 — smaller, faster, and significantly better on the two things hardest for OCR models to get right: complex tables and multilingual. Live on the Datalab API today. Details ↓

[Vikram Oberoi @voberoi]

@eatonphil Aw, man. NYC will miss you but I'm excited for you and hope you return permanently some day.

Would love to say hiya before you go.

[Vikram Oberoi @voberoi]

@harryh @tqbf You've gotta do all 35!!

[Vikram Oberoi @voberoi]

@tqbf Amazing, thank you!

[Vikram Oberoi @voberoi]

@tqbf How did you get to submit one? My wife and I keep coming up with fun ideas.

[Vikram Oberoi @voberoi]

@tqbf I found about this from this tweet and have been *hooked* all week!

And then, lo and behold: