Jacek Migdal
@jakozaur.bsky.social
Technical CEO and co-founder of Quesma. Follow for data insights. Prev 10 years at Sumo Logic (0 to $300mln ARR), Meta and NVIDIA CUDA.
Outbox pattern. @gunnarmorling.dev seems to be burnt by dual writes. For example, if you add two rows into different SQL tables in a single transaction, there are no atomic guarantees after Debezium CDC puts that into Kafka.
The Outbox pattern fixes that by adding a new table.
The Outbox pattern fixes that by adding a new table.
June 5, 2025 at 6:19 PM
Outbox pattern. @gunnarmorling.dev seems to be burnt by dual writes. For example, if you add two rows into different SQL tables in a single transaction, there are no atomic guarantees after Debezium CDC puts that into Kafka.
The Outbox pattern fixes that by adding a new table.
The Outbox pattern fixes that by adding a new table.
All you need to make founder happy is a meeting with ideal customer persona (ICP) with some sense of urgency.
January 22, 2025 at 12:40 PM
All you need to make founder happy is a meeting with ideal customer persona (ICP) with some sense of urgency.
SysOps/DevOps meetup at Allegro.
Will be speaking about Observability Query Languages.
Will be speaking about Observability Query Languages.
January 16, 2025 at 5:49 PM
SysOps/DevOps meetup at Allegro.
Will be speaking about Observability Query Languages.
Will be speaking about Observability Query Languages.
Two great summaries of 2024, databases and LLMS:
www.cs.cmu.edu/~pavlo/blog/...
simonwillison.net/2024/Dec/31/...
www.cs.cmu.edu/~pavlo/blog/...
simonwillison.net/2024/Dec/31/...
Things we learned about LLMs in 2024
A lot has happened in the world of Large Language Models over the course of 2024. Here’s a review of things we figured out about the field in the past …
simonwillison.net
January 1, 2025 at 5:49 PM
Two great summaries of 2024, databases and LLMS:
www.cs.cmu.edu/~pavlo/blog/...
simonwillison.net/2024/Dec/31/...
www.cs.cmu.edu/~pavlo/blog/...
simonwillison.net/2024/Dec/31/...
Happy new year!
Last year at Quesma we went from zero to one:
No team -> eight of us
No real code -> working product used by a few on prod
Virigin company with $2.5mln funding -> some revenue, low burn
The biggest lesson is to keep shipping while talking to users.
Last year at Quesma we went from zero to one:
No team -> eight of us
No real code -> working product used by a few on prod
Virigin company with $2.5mln funding -> some revenue, low burn
The biggest lesson is to keep shipping while talking to users.
January 1, 2025 at 12:36 PM
Happy new year!
Last year at Quesma we went from zero to one:
No team -> eight of us
No real code -> working product used by a few on prod
Virigin company with $2.5mln funding -> some revenue, low burn
The biggest lesson is to keep shipping while talking to users.
Last year at Quesma we went from zero to one:
No team -> eight of us
No real code -> working product used by a few on prod
Virigin company with $2.5mln funding -> some revenue, low burn
The biggest lesson is to keep shipping while talking to users.
All set with Advent of Code! 50 stars.
December 25, 2024 at 11:18 AM
All set with Advent of Code! 50 stars.
The penultimate day (day 24) of the Advent of Code, part 2, is tedious but doable with a tiny trick. I am glad I know a thing or two about hardware.
I am impatiently waiting for the final puzzle.
I am impatiently waiting for the final puzzle.
December 24, 2024 at 1:11 PM
The penultimate day (day 24) of the Advent of Code, part 2, is tedious but doable with a tiny trick. I am glad I know a thing or two about hardware.
I am impatiently waiting for the final puzzle.
I am impatiently waiting for the final puzzle.
One star was missing in Advent of Code, though on day 21, part 2 was somewhat hard by my standards. I have an idea how to solve it, though there are a couple of tricks. Am I getting less technical?
December 22, 2024 at 9:11 AM
One star was missing in Advent of Code, though on day 21, part 2 was somewhat hard by my standards. I have an idea how to solve it, though there are a couple of tricks. Am I getting less technical?
Back in the Advent of Code game. I’ve caught up with my delay.
December 19, 2024 at 8:08 PM
Back in the Advent of Code game. I’ve caught up with my delay.
Startups in ZIRP, 2022: Bragging how big their teams are and how fast they hire.
Startups in 2024: Bragging how small their team is.
E.g. Quesma is in that camp.
Converse job ad:
Startups in 2024: Bragging how small their team is.
E.g. Quesma is in that camp.
Converse job ad:
December 17, 2024 at 10:34 AM
Startups in ZIRP, 2022: Bragging how big their teams are and how fast they hire.
Startups in 2024: Bragging how small their team is.
E.g. Quesma is in that camp.
Converse job ad:
Startups in 2024: Bragging how small their team is.
E.g. Quesma is in that camp.
Converse job ad:
In startups, the highs can be incredibly rewarding, while the lows can be deeply challenging, but both are part of the experience.
December 16, 2024 at 5:35 PM
In startups, the highs can be incredibly rewarding, while the lows can be deeply challenging, but both are part of the experience.
Day 14, part 2 of the #AdventOfCode, is my favourite so far. In changing ASCII art, find a Christmas tree. I love that spec; it matches the reality.
December 14, 2024 at 11:49 AM
Day 14, part 2 of the #AdventOfCode, is my favourite so far. In changing ASCII art, find a Christmas tree. I love that spec; it matches the reality.
Halfway through #AdventOfCode 2024, 22 stars so far: adventofcode.com/2024/day/11
Day 11 - Advent of Code 2024
adventofcode.com
December 11, 2024 at 7:14 PM
Halfway through #AdventOfCode 2024, 22 stars so far: adventofcode.com/2024/day/11
Native support for Iceberg has been announced in S3:
aws.amazon.com/s3/features/...
aws.amazon.com/s3/features/...
Tabular Data Storage At Scale - Amazon S3 Tables - AWS
Amazon S3 Tables deliver an efficient way way to store tabular data at scale, helping you optimize query performance and costs as your data lake grows.
aws.amazon.com
December 3, 2024 at 9:38 PM
Native support for Iceberg has been announced in S3:
aws.amazon.com/s3/features/...
aws.amazon.com/s3/features/...
Besides generating social media buzz, there is rarely an upside to being obsessed about past mishaps.
Life is short. Move on. The next thing can be huge. No point in wasting your neural net cycles on that; you need all of them for the next big thing.
Life is short. Move on. The next thing can be huge. No point in wasting your neural net cycles on that; you need all of them for the next big thing.
December 3, 2024 at 8:34 PM
Besides generating social media buzz, there is rarely an upside to being obsessed about past mishaps.
Life is short. Move on. The next thing can be huge. No point in wasting your neural net cycles on that; you need all of them for the next big thing.
Life is short. Move on. The next thing can be huge. No point in wasting your neural net cycles on that; you need all of them for the next big thing.
AI saved my day when I was exporting invoices from accounting system. Still I get best results when I micro-manage AI aggresively. Any time I trust AI too much, it ends up poorly.
December 2, 2024 at 6:45 PM
AI saved my day when I was exporting invoices from accounting system. Still I get best results when I micro-manage AI aggresively. Any time I trust AI too much, it ends up poorly.
I just completed "Historian Hysteria" - Day 1 - Advent of Code 2024 #AdventOfCode adventofcode.com/2024/day/1
December 1, 2024 at 6:26 PM
I just completed "Historian Hysteria" - Day 1 - Advent of Code 2024 #AdventOfCode adventofcode.com/2024/day/1
In the spirit of Thanksgiving engineering teams across the globe should do a thankfulness exercise.
During daily or status each member says what they are thankful for. Simple but works.
During daily or status each member says what they are thankful for. Simple but works.
November 29, 2024 at 5:46 PM
In the spirit of Thanksgiving engineering teams across the globe should do a thankfulness exercise.
During daily or status each member says what they are thankful for. Simple but works.
During daily or status each member says what they are thankful for. Simple but works.
Most PMs hate refactoring, but our PM at Quesma asked for that.
Why?
1. Good overall velocity.
2. Paid clients, so we have to graduate from early-startup mode.
3. One architecture that would cover more use cases.
Why?
1. Good overall velocity.
2. Paid clients, so we have to graduate from early-startup mode.
3. One architecture that would cover more use cases.
November 27, 2024 at 3:36 PM
Most PMs hate refactoring, but our PM at Quesma asked for that.
Why?
1. Good overall velocity.
2. Paid clients, so we have to graduate from early-startup mode.
3. One architecture that would cover more use cases.
Why?
1. Good overall velocity.
2. Paid clients, so we have to graduate from early-startup mode.
3. One architecture that would cover more use cases.
Full room at DSS conference for my talk about data trends in 2024.
November 22, 2024 at 3:16 PM
Full room at DSS conference for my talk about data trends in 2024.
Adam Szymanski, CTO & Founder of Oxla at Data Science Summit Main Stage.
November 22, 2024 at 10:23 AM
Adam Szymanski, CTO & Founder of Oxla at Data Science Summit Main Stage.
“Based on what you know about me, draw a picture of what you think my life looks like.”
ChatGPT
ChatGPT
November 22, 2024 at 8:44 AM
“Based on what you know about me, draw a picture of what you think my life looks like.”
ChatGPT
ChatGPT
Speaking tomorrow at Data Science Summit: What’s New in Data Platforms for 2024?
November 21, 2024 at 6:38 PM
Speaking tomorrow at Data Science Summit: What’s New in Data Platforms for 2024?
Time to come home from a trip to the USA.
November 17, 2024 at 11:43 PM
Time to come home from a trip to the USA.