Software Engineer. Bringing Observability for financial data and business metrics to the innovation economy @ StandardMetrics. Opinions are my own.
Guadalajara, Jalisco
Joined December 2008
- Tweets23.9K
- Following3K
- Followers1.1K
- Likes7.3K
Eduardo Romero retweeted
"My colleague pushed code that doesn't match what I had in mind."
How bad is that really? Consider: it's now much faster and easier to change code.
Years ago a wrong decision could cause days and weeks of refactoring.
Now? Probably less. Adjust.
Eduardo Romero retweeted
One of the trickier parts of building UK stock market pipelines is the poor standardization.
I found that the concept of “revenue” varies a lot for each company.
For example, HSBC files "total income" under a tag it made up itself.
Lloyds tagged its "other income" line as Revenue, £5.8B, a fraction what it earns.
Aviva's Revenue tag held fee income alone, £1.5B, when its insurance revenue is £25B.
For serious research, your agent can’t just read filings and map numbers on the fly.
It’s better to use a deterministic API that’s mapped the tricky edge cases correctly.
Eduardo Romero retweeted
Built by the Electric team (who also did pglite, i.e. Postgres on WASM), Neon Realtime is an incremental query engine to make it dramatically easier to build apps that sync data in realtime.
It leverages the Lakebase architecture's separation of storage and compute to run on a separate compute instance. This avoids overwhelming the primary write Postgres instance and pushes the write operations incrementally to all the subscribers.
Check it out. You can now build complex realtime sync apps in just a few lines of code. This is a great example of how hardcore database technologies can make the app developers' lives much simpler.
Eduardo Romero retweeted
Pydantic AI now supports Gemini 4 Argon, with long decode.
One Argon request stops at 32K output tokens (64K streamed). Set a bigger output budget and Pydantic AI follows Google's continuation token for you: one agent run, one combined response.
pydantic.io/ifgZJ
Eduardo Romero retweeted
If I bought a new laptop today, I don't think I would install any local dev tools / languages at all. I’d push all my dev work, open source included, to cloud agents.
I think I could do everything I currently do with nothing but Laravel Cloud and @AmpCode Orbs.
Going to try to go cloud-only this week.
Eduardo Romero retweeted
Should you waste tokens having your coding agent use grep, or hand the search to a tiny local model?
We release ColGrep-Agent-2B! A code-search agent that runs locally, searches with ColGrep, reads files, & returns locations + code to your main agent
-> 34% fewer API tokens
Eduardo Romero retweeted
Agent reading a 10M-line log? grep ERROR gives it 398,810 lines. Try templates:
chdb.query("""
SELECT replaceRegexpAll(line,'[0-9a-f]{8,}|[0-9]+','#') t, count() n
FROM file('app.log',LineAsString)
WHERE line LIKE '%ERROR%'
GROUP BY t ORDER BY n DESC LIMIT 5""")
5 rows, 0.2s.
Eduardo Romero retweeted
Python 3.15 is out now.
Here's our overview of what's new, including lazy imports, a new statistical sampling profiler (Tachyon), sentinel types and more.
astral.sh/blog/python-3.15
Eduardo Romero retweeted
Agent Mode in the DuckDB CLI 🤖 🦆
The #DuckDB v2.0 CLI has an agent mode that gets AI coding agents to a correct result faster and with fewer tokens.
When the CLI detects an agent, it prints compact Markdown tables instead of padded boxes, says clearly when a result was cut, stops runaway queries early, and reports errors as JSON. Long queries announce their expected cost before they start, so the agent can decide whether to wait.
Learn more in today’s blog post: duckdb.org/2026/10/09/agent-…
Eduardo Romero retweeted
Our old positioning on our website was around AI-driven portfolio management.
That's an accurate description for what we do, but we've chosen something bigger and bolder to represent what we're building:
Portfolio superintelligence ⚡
Eduardo Romero retweeted
We call it 'Hypertenancy' because that's cooler and easier than saying 'per-customer compute isolation' all the time.
It means every user gets their own duckling, a dedicated DuckDB instance with its own compute. This is what MotherDuck customers building analytics into their apps get out of it 👇
🦆 No noisy neighbors. One user's monster scan can't slow down anyone else's dashboard.
🦆 Size each user on their own. Give your heaviest users a bigger duckling without paying to scale a whole cluster.
🦆 A sandbox for every agent. A runaway query stays inside its own duckling, so it can't blow up your bill or take down your warehouse.
🦆 Cost per user. With every user on their own duckling, you can see exactly what each one costs you.
🪿 Pay for what's running. Ducklings spin up on the first query and scale to zero when idle.
How it works 👇
motherduck.com/product/hyper…
Eduardo Romero retweeted
Spice Cayenne with @vortexdotdev picks a compression scheme per column, from the data's own stats:
- FastLanes: SIMD bit-packed integers
- ALP: lossless floats
- FSST: strings with O(1) random access
- dictionary, delta, run-end, nested as needed
spice.ai/blog/vortex-at-spic…
Eduardo Romero retweeted
SQLAlchemy 2.1 was released 2 weeks ago, it adds several typing improvements and has @psycopg's latest version as the default for Postgres 🎉
SQLModel 0.0.48, just released, adds support for it, so you can get the new cool stuff by default. 🥳
sqlmodel.tiangolo.com
Eduardo Romero retweeted
How can you trust probabilistic AI in a high stakes domain like customs compliance?
At @flexport our answer is to use the LLMs to write code, which once signed off on by human experts runs deterministically every time.
I explain how we're using this approach in our new business rules engine for handling customs compliance automation.
Eduardo Romero retweeted
We just released Polars 2.0.
It removed many of our legacy decisions makes the streaming engine our default and promotes SQL to a first class citizen within Polars.
It comes with initial out-of-core (spill to disk) support, a new Map data type and a lot of performance improvements. In fact, we think Polars is now one of the fastest analytical SQL engines on a single node. See benchmarks in the post: pola.rs/posts/release-polars…
Eduardo Romero retweeted
Moving from individual proofs to large-scale autoformalization requires new tools. We introduce Choir, an open protocol for multi-agent autoformalization that decomposes projects into tasks, coordinates contributions through GitHub, and deterministically checks them before merging.
Choir is modular, open source, and supports Lean 4, Isabelle, and Rocq.
Repository: github.com/Weber-GeoML/Choir
Preprint: arxiv.org/abs/2609.31903
Led by @yidi_qi. Supported by @darpa expMath.