Matt
@mattzcarey.com
ex pro windsurfer 🇲🇹 🌊 AI engineer. I work on streamlining access to large data sources, agents and retrieval - other interests in mech interp and data efficient fine-tuning.
Super fun to speak at a packed ai engineer london. My talk was about how we eval agentic RAG with Langsmith
Thanks @robbiehudson.bsky.social and crew for having me and for organising a great event 🙏
Thanks @robbiehudson.bsky.social and crew for having me and for organising a great event 🙏
January 16, 2025 at 11:07 PM
Super fun to speak at a packed ai engineer london. My talk was about how we eval agentic RAG with Langsmith
Thanks @robbiehudson.bsky.social and crew for having me and for organising a great event 🙏
Thanks @robbiehudson.bsky.social and crew for having me and for organising a great event 🙏
We have been building something for a while at StackOne with no idea what it should be called.
Google just told us.
We are building an extension
an extension for the whole HR tech ecosystem.
Google just told us.
We are building an extension
an extension for the whole HR tech ecosystem.
January 13, 2025 at 12:23 PM
We have been building something for a while at StackOne with no idea what it should be called.
Google just told us.
We are building an extension
an extension for the whole HR tech ecosystem.
Google just told us.
We are building an extension
an extension for the whole HR tech ecosystem.
i've been building a thing 🦓
the api is coming together.
the model compiled by zml
I think this leans well on zml pros.
One model, many backends, zero code change.
the api is coming together.
the model compiled by zml
I think this leans well on zml pros.
One model, many backends, zero code change.
January 13, 2025 at 12:21 PM
i've been building a thing 🦓
the api is coming together.
the model compiled by zml
I think this leans well on zml pros.
One model, many backends, zero code change.
the api is coming together.
the model compiled by zml
I think this leans well on zml pros.
One model, many backends, zero code change.
London looking pretty this time of year
December 30, 2024 at 4:19 PM
London looking pretty this time of year
made a lil website for my granny.
She needed some photo storage for her (soon to be an online) bookshop. Build her a nice little admin panel
and it has dark mode :)
She needed some photo storage for her (soon to be an online) bookshop. Build her a nice little admin panel
and it has dark mode :)
December 28, 2024 at 12:13 AM
made a lil website for my granny.
She needed some photo storage for her (soon to be an online) bookshop. Build her a nice little admin panel
and it has dark mode :)
She needed some photo storage for her (soon to be an online) bookshop. Build her a nice little admin panel
and it has dark mode :)
Good year. pushed pretty hard and got a lot better
December 27, 2024 at 6:36 PM
Good year. pushed pretty hard and got a lot better
Making this thing I have wanted for the last year :)
December 26, 2024 at 11:16 PM
Making this thing I have wanted for the last year :)
python is so broken but the fact that this is a thing.. js ecosystem is no better wtf
December 26, 2024 at 5:25 PM
python is so broken but the fact that this is a thing.. js ecosystem is no better wtf
Check out the latency increase by using gpt-4o instead of a specific model tag
This has to be a bug?
This has to be a bug?
December 21, 2024 at 11:08 AM
Check out the latency increase by using gpt-4o instead of a specific model tag
This has to be a bug?
This has to be a bug?
I knew gemini was the best LLM as a judge. Deepmind just proved me correct.
www.kaggle.com/facts-leader...
www.kaggle.com/facts-leader...
December 17, 2024 at 4:21 PM
I knew gemini was the best LLM as a judge. Deepmind just proved me correct.
www.kaggle.com/facts-leader...
www.kaggle.com/facts-leader...
day 11 of Advent of ML (nice to be back) and we are talking scaling at test time.
Do models reason? what is scaling at test time and will it lead to mythical AGI level reasoning?
find out more about this new scaling law:
mattzcarey.com/blog/advent-...
#blogvent #adventofml
Do models reason? what is scaling at test time and will it lead to mythical AGI level reasoning?
find out more about this new scaling law:
mattzcarey.com/blog/advent-...
#blogvent #adventofml
December 12, 2024 at 3:51 PM
day 11 of Advent of ML (nice to be back) and we are talking scaling at test time.
Do models reason? what is scaling at test time and will it lead to mythical AGI level reasoning?
find out more about this new scaling law:
mattzcarey.com/blog/advent-...
#blogvent #adventofml
Do models reason? what is scaling at test time and will it lead to mythical AGI level reasoning?
find out more about this new scaling law:
mattzcarey.com/blog/advent-...
#blogvent #adventofml
I just completed "Guard Gallivant" - Day 6 - Advent of Code 2024 #AdventOfCode adventofcode.com/2024/day/6
brute forced this so hard. Not safe or fun. Enums in zig are cool tho
brute forced this so hard. Not safe or fun. Enums in zig are cool tho
December 8, 2024 at 11:17 PM
I just completed "Guard Gallivant" - Day 6 - Advent of Code 2024 #AdventOfCode adventofcode.com/2024/day/6
brute forced this so hard. Not safe or fun. Enums in zig are cool tho
brute forced this so hard. Not safe or fun. Enums in zig are cool tho
I love when people think they know best online and end up arguing with the dude that built the actual thing.
never gets old
never gets old
December 8, 2024 at 2:18 PM
I love when people think they know best online and end up arguing with the dude that built the actual thing.
never gets old
never gets old
When the world wakes up to the power of evals. Tracing services gotta invest in their compute.
🙏 for the SREs
🙏 for the SREs
December 6, 2024 at 6:41 PM
When the world wakes up to the power of evals. Tracing services gotta invest in their compute.
🙏 for the SREs
🙏 for the SREs
Successful day. Turns out writing about best practises makes you more conscious of following your own advice.
60% still sucks but getting there.
60% still sucks but getting there.
December 5, 2024 at 6:15 PM
Successful day. Turns out writing about best practises makes you more conscious of following your own advice.
60% still sucks but getting there.
60% still sucks but getting there.
It's already day 5 of Advent of ML (also known as #blogvent) 😇
Finishing up on the mini dive into retrieval today. If you are not using these you probably should..
reranker!
An upgrade to @cohere.com rerank v3.5 bumped our internal retrieval evals 5%!!
mattzcarey.com/blog/advent-...
Finishing up on the mini dive into retrieval today. If you are not using these you probably should..
reranker!
An upgrade to @cohere.com rerank v3.5 bumped our internal retrieval evals 5%!!
mattzcarey.com/blog/advent-...
December 5, 2024 at 2:55 PM
It's already day 5 of Advent of ML (also known as #blogvent) 😇
Finishing up on the mini dive into retrieval today. If you are not using these you probably should..
reranker!
An upgrade to @cohere.com rerank v3.5 bumped our internal retrieval evals 5%!!
mattzcarey.com/blog/advent-...
Finishing up on the mini dive into retrieval today. If you are not using these you probably should..
reranker!
An upgrade to @cohere.com rerank v3.5 bumped our internal retrieval evals 5%!!
mattzcarey.com/blog/advent-...
I just completed "Ceres Search" - Day 4 - Advent of Code 2024
finally got to include a nice struct.
Also how mental is that patterns typing. Strings actually being slices of unsigned integers never get old.
#AdventOfCode adventofcode.com/2024/day/4
finally got to include a nice struct.
Also how mental is that patterns typing. Strings actually being slices of unsigned integers never get old.
#AdventOfCode adventofcode.com/2024/day/4
December 4, 2024 at 10:06 PM
I just completed "Ceres Search" - Day 4 - Advent of Code 2024
finally got to include a nice struct.
Also how mental is that patterns typing. Strings actually being slices of unsigned integers never get old.
#AdventOfCode adventofcode.com/2024/day/4
finally got to include a nice struct.
Also how mental is that patterns typing. Strings actually being slices of unsigned integers never get old.
#AdventOfCode adventofcode.com/2024/day/4
we have bluesky enabled comments on my new latest Advent of ML blog post and forever more.
you can just do stuff :))))))))
mattzcarey.com/blog/advent-...
you can just do stuff :))))))))
mattzcarey.com/blog/advent-...
December 3, 2024 at 11:37 PM
we have bluesky enabled comments on my new latest Advent of ML blog post and forever more.
you can just do stuff :))))))))
mattzcarey.com/blog/advent-...
you can just do stuff :))))))))
mattzcarey.com/blog/advent-...
I just completed "Mull It Over" - Day 3 - Advent of Code 2024
Pretty grim but it works, zig feeling more natural
Needle in a haystack indexOf is so cool and feels v natural.
#AdventOfCode adventofcode.com/2024/day/3
Pretty grim but it works, zig feeling more natural
Needle in a haystack indexOf is so cool and feels v natural.
#AdventOfCode adventofcode.com/2024/day/3
December 3, 2024 at 11:09 PM
I just completed "Mull It Over" - Day 3 - Advent of Code 2024
Pretty grim but it works, zig feeling more natural
Needle in a haystack indexOf is so cool and feels v natural.
#AdventOfCode adventofcode.com/2024/day/3
Pretty grim but it works, zig feeling more natural
Needle in a haystack indexOf is so cool and feels v natural.
#AdventOfCode adventofcode.com/2024/day/3
good days work.
big increase on evals.
where's my bonus?
big increase on evals.
where's my bonus?
December 3, 2024 at 3:32 PM
good days work.
big increase on evals.
where's my bonus?
big increase on evals.
where's my bonus?
My ml engineer brain tells me this is just a tokenizer issue.
My comp sci brain tells me the tokenizers are fundamentally the wrong abstraction and this is such a hack. Which is right?
My comp sci brain tells me the tokenizers are fundamentally the wrong abstraction and this is such a hack. Which is right?
December 2, 2024 at 4:02 PM
My ml engineer brain tells me this is just a tokenizer issue.
My comp sci brain tells me the tokenizers are fundamentally the wrong abstraction and this is such a hack. Which is right?
My comp sci brain tells me the tokenizers are fundamentally the wrong abstraction and this is such a hack. Which is right?
Day 2 - Advent of Code 2024 done
still getting to grips with zig std lib. Been using ArrayLists over comptime + slices which I think is bad form - c'est la vie
#AdventOfCode adventofcode.com/2024/day/2
still getting to grips with zig std lib. Been using ArrayLists over comptime + slices which I think is bad form - c'est la vie
#AdventOfCode adventofcode.com/2024/day/2
December 2, 2024 at 1:04 PM
Day 2 - Advent of Code 2024 done
still getting to grips with zig std lib. Been using ArrayLists over comptime + slices which I think is bad form - c'est la vie
#AdventOfCode adventofcode.com/2024/day/2
still getting to grips with zig std lib. Been using ArrayLists over comptime + slices which I think is bad form - c'est la vie
#AdventOfCode adventofcode.com/2024/day/2