Omar Kilani Profile Banner
Omar Kilani Profile
Omar Kilani

@omarkilani

998
Followers
759
Following
23
Media
530
Statuses

eng @groqinc , co-founder @rememberthemilk . @waymo fan account. you can just do things enthusiast.

San Francisco, CA
Joined March 2007
Don't wanna be here? Send us removal request.
Pinned Tweet
@omarkilani
Omar Kilani
18 years
Hello, world.
2
0
3
@omarkilani
Omar Kilani
4 months
400t/s L3 70b has been achieved internally.
11
12
168
@omarkilani
Omar Kilani
2 months
Huge day at @GroqInc ! 🚀 Our world-class engineering team has been relentlessly advancing the field of AI inference. Today, their hard work pays off as we secure $640M in funding. Massive kudos to the team! 🫡
@Techmeme
Techmeme
3 months
AI chip startup Groq raised a $640M Series D led by BlackRock at a $2.8B valuation, up from $1B after raising $300M in 2021, and adds an Intel executive as COO ( @vandermey / Bloomberg) 📫 Subscribe:
8
34
237
0
6
66
@omarkilani
Omar Kilani
5 months
The @GroqInc compiler team are all literal geniuses. We improved Mixtral 8x7 t/s/u by an entire GPT-4o t/s/u (474 to 585 median) with compiler improvements. Just getting started. 🫡
@GavinSherry
Gavin
5 months
We just pushed another optimization to @MistralAI Mixtral 8x7 to @GroqInc . Users will see a ~20% throughput improvement 🙌. These enhancements are driven by compiler software team’s relentless focus on throughput and latency.
Tweet media one
7
7
66
4
11
63
@omarkilani
Omar Kilani
5 months
L3 8B at @GroqInc , a timeline: Last Friday: 946t/s Today at 3:19p: 1157t/s Today at 5:44p: 1270t/s Somehow, though, this is just the start.
@GavinSherry
Gavin
5 months
Good suggestion by pundits to run on fewer chips. Turns out that made @GroqInc faster 🏎️🏎️🏎️
Tweet media one
2
7
33
2
5
42
@omarkilani
Omar Kilani
5 months
@yacineMTB We made huge latency improvements across the board this week. More to come. 🫡
2
0
38
@omarkilani
Omar Kilani
5 months
Casually shipping 348t/s of 70B goodness on a Friday afternoon.
@GavinSherry
Gavin
5 months
The @GroqInc team just shipped some optimizations pushing per user tokens per second higher for @AIatMeta Llama 3 70b. Looking forward to seeing what everyone builds this weekend.
Tweet media one
3
6
46
2
3
36
@omarkilani
Omar Kilani
3 months
@jiayq @GroqInc We really appreciate the kind words! 🫡
1
2
29
@omarkilani
Omar Kilani
5 months
Tokens go burrr 😍😍😍
@GroqInc
Groq Inc
5 months
We're working hard to deploy more GroqRacks to serve the dev community's growing demand! 🚀
15
21
232
1
1
23
@omarkilani
Omar Kilani
5 months
@CloudflareDev
Cloudflare Developers
5 months
AI gateway now supports @GroqInc and @Cohere ! Unleash the full potential of your language model, no matter where you are. 👂 We're all ears - let us know which providers or features you'd like to see next!
Tweet media one
2
18
109
0
2
22
@omarkilani
Omar Kilani
2 months
We at @GroqInc are thrilled to have the ridiculously talented @AarushSah_ join us full time — Aarush just celebrated his 18th birthday. 🤯
@AarushSah_
Aarush Sah
2 months
Internship got cut a little short. Happy to share that I'm now full-time at @GroqInc - LET'S COOK
Tweet media one
50
16
930
1
0
21
@omarkilani
Omar Kilani
5 months
Come make @GroqInc even faster and better. We’re not stopping until we hit 0ms TTFT.
@GavinSherry
Gavin
5 months
We’re expanding the @GroqInc team. If you’re a no nonsense engineer able to do deep systems work with great intensity, DM me.
6
3
38
0
2
19
@omarkilani
Omar Kilani
5 months
L3 8B in production running at 1157t/s/u with the full 8k context window. Only at @GroqInc . 🫡
@sundeep
sunny madra
5 months
The beauty of the @GroqInc design is that it can always be faster... #thefastestinference on, 14nm technology 😉
Tweet media one
6
3
45
0
2
19
@omarkilani
Omar Kilani
3 months
. @GroqInc is making agentic workloads a reality— at Groq speed. Huge ship from @RickLamers 🫡
@RickLamers
Rick Lamers
3 months
I’ve been leading a secret project for months … and the word is finally out! 🛠️ I'm proud to announce the Llama 3 Groq Tool Use 8B and 70B models 🔥 An open source Tool Use full finetune of Llama 3 that reaches the #1 position on BFCL beating all other models, including
Tweet media one
74
236
1K
1
2
18
@omarkilani
Omar Kilani
5 months
Are you: 1. A world class software engineer. 2. Obsessed with performance optimization. 3. Into helping build the world's fastest inference engine. Nice. 🫡 @GroqInc is hiring distributed systems engineers: Join us on our quest to 0ms TTFT.
1
1
16
@omarkilani
Omar Kilani
4 months
👀
Tweet media one
0
1
17
@omarkilani
Omar Kilani
5 months
If you feel like this is still too high — me too. We’re hiring distributed systems engineers. DM for info. 🫡
@sundeep
sunny madra
5 months
Latency 😉😜😘 @GroqInc
Tweet media one
4
6
39
1
1
14
@omarkilani
Omar Kilani
5 months
🫡
@ArtificialAnlys
Artificial Analysis
5 months
Groq extends its lead and is serving Llama 3 8B at almost 1,200 output tokens/s! @GroqInc 's Llama 3 8B speed improvements seen in their chat interface we can now confirm are reflected in performance of their API. This represents the fastest language model inference performance
Tweet media one
5
11
62
1
2
13
@omarkilani
Omar Kilani
4 months
We cranked the input speed on this one to 11, thanks to the ingenuity of the @GroqInc compiler team. 🫡
Tweet media one
@ArtificialAnlys
Artificial Analysis
4 months
Fast to launch & very fast output speed! Groq has launched their Gemma 2 9B offering and is serving it at ~600 output tokens/s Gemma 2 9B is worthy alternative to Llama 3 8B and other smaller models. It is particularly attractive for generalist and communication-focused
Tweet media one
4
22
70
0
3
14
@omarkilani
Omar Kilani
9 months
My guy @ianlandsman laying out the reasons why Section 174 is an extinction level event for a lot of small software companies.
@ianlandsman
Ian Landsman
9 months
I have a few quotes about 174 in Politico's Morning Tech today.
Tweet media one
2
4
29
0
2
14
@omarkilani
Omar Kilani
6 months
New @Waymo map + “Ambient Vibes” … awesome.
Tweet media one
1
0
12
@omarkilani
Omar Kilani
6 months
@legolasyiu This isn’t even our final form. :)
2
1
10
@omarkilani
Omar Kilani
5 months
Rick shipped a thing: streaming tool calls @GroqInc . 🚢 Always deploy on Fridays (trying to get Rick to YOLO more often).
@RickLamers
Rick Lamers
5 months
I shipped a thing! On Friday, haha yes I’m crazy
7
3
51
1
2
10
@omarkilani
Omar Kilani
5 months
“Invented” real time username availability checks on signup forms.
@BigMeanInternet
Malcolm Harris
5 months
What's something in the world you know you're directly responsible for but if you were to claim credit you'd sound crazy?
995
251
4K
2
1
9
@omarkilani
Omar Kilani
5 months
A lot of very late nights went into this but those TTFT numbers are 😍. More to come.
@sundeep
sunny madra
5 months
The fastest . Ai ⚡️⚡️⚡️
Tweet media one
Tweet media two
9
9
71
2
2
9
@omarkilani
Omar Kilani
8 months
The fate of small software companies hanging in the balance on this vote…
@elwasson
Erik Wasson
8 months
TAX: @SenSchumer tells me he intends to bring business / child tax bill to the floor for vote
2
21
72
0
3
7
@omarkilani
Omar Kilani
9 months
It really sucks that thousands of software companies and their owners, employees, etc are beholden to the dumbest people imaginable, but that’s where we are I guess. H.R. 7024 is the easiest win-win-win to come out of Congress in 2 years. Just pass it.
@burgessev
Burgess Everett
9 months
News: Senate Minority Whip Thune says Senate Republicans will block the House-passes tax deal without an opportunity to amend it on the floor or in committee. Says GOP wants changes to child tax credit work requirements
25
54
110
0
0
6
@omarkilani
Omar Kilani
4 months
Join us on our quest to make all these numbers better every day:
@sundeep
sunny madra
4 months
Our engineering team is cooking. Latency, Throughput, and Quality all keep improving! Across different models.
Tweet media one
Tweet media two
Tweet media three
10
19
95
0
3
9
@omarkilani
Omar Kilani
9 months
Section 174 is one step closer to getting fixed. 🎉
@RichardRubinDC
Richard Rubin
9 months
And it’s done. Tax bill passes 357-70.
6
14
93
0
1
8
@omarkilani
Omar Kilani
4 months
@swyx [insert disclaimer about parameters, variance, non-determinism, macro averages, MoE, etc etc here]
Tweet media one
Tweet media two
0
1
8
@omarkilani
Omar Kilani
7 months
@ianlandsman For any given question, the answer is always: “just use Postgres”.
1
1
8
@omarkilani
Omar Kilani
10 months
Have travelled over 1,000 miles in a @Waymo A mind blowing, magical feat of engineering and one of the most important advances in tech in a long time.
Tweet media one
@Waymo
Waymo
10 months
Thanks to our Waymo One riders for riding with us more than ever before this year. We can't wait for what's to come in 2024!
Tweet media one
Tweet media two
Tweet media three
Tweet media four
5
27
124
2
0
7
@omarkilani
Omar Kilani
5 months
@rudiranck @ArtificialAnlys @ylecun @GroqInc Just circling back on this one. 🫡
Tweet media one
Tweet media two
0
1
7
@omarkilani
Omar Kilani
7 months
Tweet media one
0
0
7
@omarkilani
Omar Kilani
5 months
@MingXDynasty @GroqInc Awesome. IMHO, competition is great, and it’s nice to see what the H200 can do at scale. OpenAI should simply run GPT on the LPU. :) (The exciting thing about this is that we’re still super early in the LPU performance story.)
2
1
7
@omarkilani
Omar Kilani
6 months
3
1
7
@omarkilani
Omar Kilani
9 months
@samcraigjohnson 99% of SaaS co’s can get to 10M+ users on a single Scale-A5 from @OVHcloud_US for $663/m (maybe get a couple for redundancy, Postgres, etc).
Tweet media one
0
0
7
@omarkilani
Omar Kilani
5 months
We’re doing an AMA! 🫡 Stop by and ask your Groq Q’s. 🤔
@GroqInc
Groq Inc
5 months
Want to know how Groq can scale to accommodate the growing demand for inference and how the scaling limitations of traditional legacy architectures can be overcome? Tune in on June 5 to find out at our upcoming AMA.
Tweet media one
0
3
23
0
1
6
@omarkilani
Omar Kilani
3 months
🫡
@AIatMeta
AI at Meta
3 months
@GroqInc Impressive work from the Groq team! 👏
2
8
61
0
0
7
@omarkilani
Omar Kilani
3 months
LFG 🫡
@AIatMeta
AI at Meta
3 months
Starting today, open source is leading the way. Introducing Llama 3.1: Our most capable models yet. Today we’re releasing a collection of new Llama 3.1 models including our long awaited 405B. These models deliver improved reasoning capabilities, a larger 128K token context
284
1K
6K
0
0
6
@omarkilani
Omar Kilani
11 months
@golang Happy birthday! 🎉
0
0
1
@omarkilani
Omar Kilani
6 months
@koltregaskes @GroqInc Coming very soon! (The playground does support it in the meantime.)
1
1
6
@omarkilani
Omar Kilani
6 months
@yacineMTB We can make this happen.
0
0
6
@omarkilani
Omar Kilani
4 months
Went on a walk…
Tweet media one
1
0
5
@omarkilani
Omar Kilani
2 months
0
0
4
@omarkilani
Omar Kilani
4 months
@brianwilt Common Waymo enthusiast knowledge tbh.
0
0
4
@omarkilani
Omar Kilani
5 months
🫡
@LauraModiano
Laura Modiano
5 months
The winner of the Paris @cerebral_valley hackathon was definitely @GroqInc Most requested and seen on screen
Tweet media one
Tweet media two
Tweet media three
1
8
43
0
0
4
@omarkilani
Omar Kilani
9 months
Promising signs for Section 174 (if the Senate does their job).
@VP
Vice President Kamala Harris
9 months
Good news: The Child Tax Credit bill is headed to the Senate. While @POTUS and I continue to fight for the full expanded Child Tax Credit, this bill should be passed quickly. President Biden is ready to sign it into law.
734
2K
6K
0
0
3
@omarkilani
Omar Kilani
8 years
@ianlandsman just gonna leave this here for you
Tweet media one
0
3
4
@omarkilani
Omar Kilani
5 months
That bat man is incredible @geeksplainer (Multi modal is coming to @GroqInc :)
@sundeep
sunny madra
5 months
preview: multimodal model on @GroqInc 👀👀👀
22
21
227
0
0
3
@omarkilani
Omar Kilani
9 months
@CFDevelop I agree, but the issue (IMHO) is that ~5% of people have the self motivation to work like that.
1
0
3
@omarkilani
Omar Kilani
10 months
Can’t believe this hasn’t been fixed yet. Everyone just keeps hoping Congress fixes this and we’re already at Dec 13. Literal death sentence for most small tech companies. Even worse for LLCs and S-Corps as you personally generate highly inflated phantom income that gets taxed.
1
1
4
@omarkilani
Omar Kilani
9 months
@brianwilt My Waymo drove through the smoke and flames of a car that was on fire yesterday. Took like 2 seconds to think/wait for a lull in oncoming traffic and go around. Was awesome. Wonder if you guys had that in the simulator…
Tweet media one
Tweet media two
1
0
3
@omarkilani
Omar Kilani
9 months
Section 174 almost fixed (international expenses still broken though, but at least the politics of that make weird sense).
@BrendanPedersen
Brendan Pedersen
9 months
Longer statement just now from the White House on the bipartisan tax deal, via spox Michael Kikukawa:
Tweet media one
1
10
13
0
0
3
@omarkilani
Omar Kilani
4 months
@mycoliza tbf the Google GLB is black magic unrivaled by anything that exists elsewhere.
0
0
3
@omarkilani
Omar Kilani
9 months
There are two types of engineers…
@tekbog
terminally onλine εngineer 🇺🇦
9 months
it's just sending one json from one service to another how hard can it be?
Tweet media one
64
295
4K
0
0
2
@omarkilani
Omar Kilani
5 months
@jmduke I did this once. It took 5 years and cost $5m. It was successful in the end, but it was also the worst thing I’ve ever done software wise.
0
0
2
@omarkilani
Omar Kilani
5 months
@eyeofenceladus Honestly… it’s amazing.
0
0
2
@omarkilani
Omar Kilani
3 months
@RickLamers Simply write the least amount of code possible in the first place. :)
1
0
3
@omarkilani
Omar Kilani
1 year
@brianwilt @Waymo True. I get extremely car sick in (most?) other cars but never in Waymo.
0
0
3
@omarkilani
Omar Kilani
6 months
@HarleyW_Alt @GroqInc It’s coming soon! (The playground has dark mode already, if you’d rather not stare at the sun.)
0
0
2
@omarkilani
Omar Kilani
5 months
One more for today…
@sundeep
sunny madra
5 months
Fast Friday continues with speed love for @MistralAI 8x7b 👀 on @GroqInc
Tweet media one
1
1
27
0
0
3
@omarkilani
Omar Kilani
6 months
Tweet media one
1
1
3
@omarkilani
Omar Kilani
5 months
30,000 t/s in... just the start.
@sundeep
sunny madra
5 months
. @GroqInc engineers working hard on a Friday, improving our stack to get more performance out of LPUs 👀 30,000 tok/s input #youaintseennothingyet
Tweet media one
11
9
122
0
0
3
@omarkilani
Omar Kilani
6 months
@matijagrcic @rafalwilinski You can put a sleep() in between the chunks if you like. :)
0
0
3
@omarkilani
Omar Kilani
8 years
@stewart some men just want to watch the world burn.
0
0
2
@omarkilani
Omar Kilani
6 months
@RickLamers
Rick Lamers
6 months
@yar_vol @GroqInc Streaming is coming! 🔜
1
0
7
0
0
3
@omarkilani
Omar Kilani
8 months
@ianlandsman “You can just do things”
0
2
3
@omarkilani
Omar Kilani
6 months
@rickykirkendall @GroqInc Thanks for flagging this. Streaming is coming soon. :) We’ll get this limitation added to the docs in the meantime. 🫡
1
0
3
@omarkilani
Omar Kilani
5 months
0
0
2
@omarkilani
Omar Kilani
2 months
@AarushSah_ Happy birthday Aarush!!! 🎉
1
0
3
@omarkilani
Omar Kilani
8 months
Literally how all good software was made…
@jmwind
Jean-Michel Lemieux
8 months
I’ve retired from software… process. No scrum, dds, tdd, stand ups, devops, sre, micro services, retrospectives, pre and post mortems… Instead, we just build and run software together. We do use an issue tracker and a good readme. Everyone posts an eod update to our group
171
248
3K
0
1
3
@omarkilani
Omar Kilani
6 months
@MoonRotator @RickLamers is working on making it the best. :)
0
0
3
@omarkilani
Omar Kilani
8 years
@andrey_butov assuming there are ever elections again or the world still exists.
0
1
3
@omarkilani
Omar Kilani
8 months
@ptr_to_joel It’s like you were there.
1
0
3
@omarkilani
Omar Kilani
6 months
@BartronPolygon Hey Bart, meet @RickLamers who can help diagnose this. 🫡
2
0
2
@omarkilani
Omar Kilani
8 months
@PatrickFIanagan Trying to explain this to people is ridiculously frustrating.
1
0
2
@omarkilani
Omar Kilani
1 year
@nntaleb Your friend should get a Neo 2T and hook it up to Zwift: Then their cycling shoes can stay on no matter the season.
0
0
0
@omarkilani
Omar Kilani
8 years
@andrey_butov Don't watch/read the news man. I stopped years ago. Best decision ever.
1
1
2
@omarkilani
Omar Kilani
8 years
@andrey_butov @ianlandsman @dhicking 3k for an iMac... should have waited for the update that never comes... ;)
0
0
2
@omarkilani
Omar Kilani
2 years
@ianlandsman @aarondfrancis Never read the comments, Ian.
0
0
2
@omarkilani
Omar Kilani
4 months
0
0
2
@omarkilani
Omar Kilani
5 months
@BeardAintWeird_ @sundeep Hey Samee — meet @RickLamers , who’s in charge of tool calls at Groq. Feel free to reach out with more info.
1
0
2
@omarkilani
Omar Kilani
8 months
0
0
1
@omarkilani
Omar Kilani
6 months
@hive_echo This hasn’t been fully answered yet because Zuck claims L3 was designed for tool use: We (well, @RickLamers ) implemented it ourselves. Maybe you could work with Rick to see how we can improve our support.
0
0
2
@omarkilani
Omar Kilani
5 months
@felixchin1 4o is running at 109 t/s/u on a H200 which is pretty impressive, but we don’t know enough about the model to say if that’s “super fast”, IMHO. It’s very unlikely to be faster than the same model running on the LPU. I would be more than happy to spin that up for OAI. :)
1
0
2
@omarkilani
Omar Kilani
8 years
@kazuho that's pretty awesome. Good luck! :)
0
0
2
@omarkilani
Omar Kilani
2 months
0
0
2
@omarkilani
Omar Kilani
7 years
@thomasfuchs It's in a file called 'common/supplemental/supplementalData.xml' in the SVN repo under 'weekData'. Or in Java, ...
1
0
2
@omarkilani
Omar Kilani
4 years
@benostrower This tweet speaks to me on a very personal level.
0
0
2
@omarkilani
Omar Kilani
4 years
@andrey_butov @ianlandsman I’m gonna need royalties on every call to json_encode/json_decode. 🤷🏻‍♂️
0
0
2
@omarkilani
Omar Kilani
6 months
@brianwilt This legitimately sounds like the best conference ever.
0
0
1
@omarkilani
Omar Kilani
8 years
@andrey_butov it was really amazing. So was Khizr Khan.
0
0
1
@omarkilani
Omar Kilani
8 years
@andrey_butov hell is other people's code.
0
1
2