Lightnews — Scholar-powered news

Matthew Mullins

@mmullins.coginiti.co

3.9K followers 510 following 1.6K posts

technologist, philosopher, and outdoors enthusiast CTO@COGINITI

Posts Media Videos Starter Packs

Pinned

Matthew Mullins @mmullins.coginiti.co · 10d

Coming to Raleigh for All Things Open? We've got something special cooked for the data community. #databs #AllThingsOpen

Data on Tap · Luma

Join Us for a Toast to Open Data & Modern Analytics Hosted by: Confluent • Snowflake • Coginiti Pull up a pint and connect with fellow data professionals for…

luma.com

Matthew Mullins @mmullins.coginiti.co · 1d

QuackStore is a critical component of Coginiti's HybridQuery capabilities for Apache Iceberg workloads. Join me at @allthingsopen.bsky.social next week where I'll share how we're using it in production and the performance wins we're seeing.

Matthew Mullins @mmullins.coginiti.co · 1d

We've open sourced QuackStore - a block-based caching extension for DuckDB! 🦆

QuackStore dramatically speeds up queries on remote data by intelligently caching only the blocks you need.

Now available as a DuckDB community extension: github.com/coginiti-dev...

GitHub - coginiti-dev/QuackStore

Contribute to coginiti-dev/QuackStore development by creating an account on GitHub.

github.com

3 4 20

Matthew Mullins @mmullins.coginiti.co · 1d

You can now call ML models in DuckDB using the Infera extension, which is just pretty damn cool. #databs github.com/CogitatorTec...

github.com

Matthew Mullins @mmullins.coginiti.co · 6d

dbt has passed on previous attempts at acquisition from bigger players, so will be curious to see if this one happens.

Matthew Mullins @mmullins.coginiti.co · 6d

Best guess, a16z is the lead investor for both and they’re pushing a merger so the joint is big enough to IPO and exit. Combined company would have about a quarter of Informatica’s revenue but better growth potential.

1 2

Matthew Mullins @mmullins.coginiti.co · 6d

I was reflecting this morning on how we've come to be ruled by the absolute dumbest lot and how they keep winning. Depressing

1 3

Matthew Mullins @mmullins.coginiti.co · 6d

Arrow does have an associated file format, Feather, that uses the Arrow IPC file format. Not recommended for storage, but it exists.

1 1

Matthew Mullins @mmullins.coginiti.co · 6d

AI Agents don't care about UX I suppose...

Matthew Mullins @mmullins.coginiti.co · 7d

I was talking to a founder who raised $50m to get to $5m in ARR, and I couldn't tell if he was lamenting or bragging. It's wild out there

Matthew Mullins @mmullins.coginiti.co · 8d

A Stalin like moment

Matthew Mullins @mmullins.coginiti.co · 8d

You can't polish a turd...

More elegantly, analytical rigor can’t fix faulty data.

Matthew Mullins @mmullins.coginiti.co · 9d

I met someone from a Google spinout and they were upset about the moving from Google's internal infra to GCP. It's not that GCP is bad, but the internal integration is just so much better.

Matthew Mullins @mmullins.coginiti.co · 10d

Coming to Raleigh for All Things Open? We've got something special cooked for the data community. #databs #AllThingsOpen

Data on Tap · Luma

Join Us for a Toast to Open Data & Modern Analytics Hosted by: Confluent • Snowflake • Coginiti Pull up a pint and connect with fellow data professionals for…

luma.com

Matthew Mullins @mmullins.coginiti.co · 10d

As Jens Voigt would say, Shut Up Legs

Matthew Mullins @mmullins.coginiti.co · 11d

Fivetran in negotiations to purchase dbt labs, this after just acquiring SQLMesh. I’m sure they aren’t the only bidders, but should they win it’s hard to imagine they keep two data transformation frameworks in operation. Great for Coginiti to see competitors rolled up. #databs

Data Startup Fivetran In Talks to Buy Dbt Labs in Multibillion Dollar Deal

Fivetran, a startup used by companies to manage and prepare data for analytics and artificial intelligence, is in talks to buy data management companydbt Labs, according to people with direct knowledg...

www.theinformation.com

2 6

Matthew Mullins @mmullins.coginiti.co · 12d

Trump’s Supreme Court basically made it impossible to send presidents to jail

1 1

Matthew Mullins @mmullins.coginiti.co · 12d

You don't have to work to get me to see it's a terrible company

Matthew Mullins @mmullins.coginiti.co · 12d

Their corporate coffee sucks, so who really cares.

Matthew Mullins @mmullins.coginiti.co · 12d

I met the biGENiUS guys last week for the first time and thought, wow we built this a decade ago at Aginity and sold it to the likes of Bass Pro, Kroger, Best Buy, CostCo, Philip Morris, etc. I know at least a couple of those are still in operation today, though Aginity is gone.

1 1

Matthew Mullins @mmullins.coginiti.co · 12d

You can see the chicken sexing machine in action. It uses computer vision, taking four photos of the chick as it goes over the drop, to rapidly classify them as male or female. I think slightly lower accuracy than professional chicken sexers, but much higher rate of throughput.

www.targan.com

Matthew Mullins @mmullins.coginiti.co · 12d

Chicken sexing is part of a core argument for reliablism about epistemic justification. Professional chicken sexers can classify 1,000 chicks per hour at over 98% accuracy, but they barely have time to look at the near indistinguishable chicks. You can read more about reliablism and sexers in IEP

Reliabilism | Internet Encyclopedia of Philosophy

iep.utm.edu

1 1

Matthew Mullins @mmullins.coginiti.co · 12d

Great turnout for our Low-Key data meetup in Raleigh last night. Wonderful catching up with so many regulars and meeting the new people that showed up! Lots of great conversations about people, processes, careers, and the personal highlight for me... meeting someone building ml for chicken sexing!

1 1

Matthew Mullins @mmullins.coginiti.co · 13d

We want our data columnwise end to end and we want it now!

1 2

Matthew Mullins @mmullins.coginiti.co · 13d

It’s really cool to see what’s getting built using the components of the composable data stack like Apache Datafusion, Apache Arrow, and Apache Iceberg. Cloudflare’s R2SQL is just another example of what’s possible. Hey @columnar.tech where is the driver for this?

R2 SQL: a deep dive into our new distributed query engine

R2 SQL provides a built-in, serverless way to run ad-hoc analytic queries against your R2 Data Catalog. This post dives deep under the Iceberg into how we built this distributed engine, from its metad...

blog.cloudflare.com

1 1 9

Reposted by Matthew Mullins

Randy Au @randyau.com · 14d

OK!! #DataBS conf starts in 45min and I'm gonna be busy hosting, SO, things to know:

Chat is on discord, NOT zoom: bit.ly/databs-chat

Last min registration: us02web.zoom.us/webinar/regi...

Official Schedule: databsconf.com/schedule/

Join the Approaching Significance Discord Server!

Check out the Approaching Significance community on Discord - hang out with 946 other members and enjoy free voice and text chat.

bit.ly

1 10 3