One developer, about a dozen Claude Code agents around the clock: 1,652 merged PRs in a month (in the range of VS Code and CPython).
Here's what kept the codebase from falling apart, and how each number was measured:
alexmarquardt.com/ai-tools/o…
@bcherny you've described running many Claude Code sessions in parallel. This is what it took to keep a dozen of them from breaking a codebase: every failure became a rule, every rule a check. Curious which of these you've hit.
@withgraphite the comparison uses your benchmark: about 2 merged PRs per engineer per week, median PR 47 lines. By that measure the main repo's 373 a week is the output of about 186 engineers. Would be interested in your take.
"gift for grandpa" and "oranges" are both search queries — but they need completely different retrieval strategies.
The missing piece? A governance layer between the user and the engine.
Part 1 of 7: elastic.co/search-labs/blog/…
@elastic @Elasticsearch
Single-pass AI code review spreads itself thin. So I built a tool that runs Claude Code in an autonomous loop — multiple passes, each building on the previous one's cleanup.
Point it at a codebase, walk away, come back to better code.
github.com/alexander-marquar…
@AnthropicAI
Credible e-commerce search demos need realistic catalogs: working images, usable categories/attrs, and enough variety to show query rewriting / re-ranking effects.
alexmarquardt.com/elastic/ec…
Over the years, I’ve written about search and distributed systems. I recently reorganized my personal site, alexmarquardt.com/, to make recurring architectural patterns easier to navigate.
Additive business boosts in search often break relevance at scale. Multiplicative influence is a more stable approach.
elastic.co/search-labs/blog/…
Learn about the impact of high replica counts in Elasticsearch, and how to ensure cluster stability by right-sizing your replicas. elastic.co/search-labs/blog/…
Here is a new article about Elasticsearch! elastic.co/search-labs/blog/…
#elastic #elasticsearch @elastic
Here is a new blog post about Logstash!
elastic.co/blog/production-d…
#logstash #elasticsearch @elastic
Alexander Marquardt retweeted
In this article, we discuss the fundamental architectural difference that makes data warehouses appropriate for analytics, and that makes operational databases appropriate for operational workloads. airbyte.com/blog/data-wareho…
Alexander Marquardt retweeted
Improve your GDPR compliance with an EtLT (Extract, “tweak”, Load, Transform) approach to data integration – in which a very light “tweak” (small “t”) transformation is done on the data before it reaches the destination.
airbyte.com/blog/etlt-gdpr-c…
Alexander Marquardt retweeted
Learn about Airbyte's ELT implementation, explore the SQL code used under-the-hood for full refresh synchronizations, and see how your replicated data looks
airbyte.com/tutorials/full-d…
Alexander Marquardt retweeted
Data integration can be used to create a unified view of your organization's data. In this article we discuss the five key questions you need to consider for your data integration strategy.
airbyte.com/blog/data-integr…
My most recent blog discusses how you may use Logstash to help make your data more GDPR compliant. alexmarquardt.com/2022/01/20… #logstash #elasticsearch
Alexander Marquardt retweeted
AWS sees itself as a friend to the open source community / industry.
How the industry sees @awscloud:
My article on stemmers and synonyms is now live on Elastic's blog - elastic.co/blog/improve-sear…
@elastic #elasticsearch
