Now
as of
- Running Hagen, the AI agent that drafts the first root-cause analysis for YugabyteDB support, in production. How it works, YFTT 162 →
- Keeping an archive of European electricity data, five countries so far, that refreshes itself every day and re-checks the trailing ninety for restatements. A day on the grid →
- Tracking a pre-registered test of the Portuguese season as it plays out, against a prediction sealed before kick-off. The season as it stands →
Everything, by date
RSSStudies published here, and work reviewed or hosted elsewhere, in one list. A talk is labelled as a talk: it carries no external review and is not listed as though it did.
- StudyFour frontier models on the same support queue: what separates them, and what does not
A year of these rounds has taught me that most model comparisons are run on the work a system already does well. Ours was too. On the tickets our Support team closed without help, four frontier models are interchangeable: they name the root cause between 87% and 96% of the time and you could pick on price alone.
- StudyPutting an agent on the incident queue: what I measured before I trusted it
I run Applied AI and Support at Yugabyte. Both halves of that title report to the same outcome: how well, and how soon, we help a customer who needs us. So when we put an AI agent in front of the support engineers, the question was never "which model scores highest". It was: can an engineer act on what this agent says, how often will it cost them time, and can we tell in advance.
- LiveThe Portuguese season as it stands
This page changes. Every other report here is dated and finished. This one tracks a season in progress against a test that was sealed before the season started, and it will keep changing until that test resolves on 15 November 2026.
- StudyHow much of a booking index is real?
A club's yellows-per-foul rate over one season is mostly noise. 15% of the spread between clubs is a real club property; the other 85% is the Poisson accident of one season's bookings.
- StudyThe lever nobody pulls
Whether a foul is punished depends enormously on where and when it happens. A foul in your own defensive fifth is carded 31.4% of the time; the same offence in the attacking fifth, 8.9%. A foul in the opening quarter of an hour is carded 6.5% of the time; after ninety minutes, 25.6%.
- StudyA club that fouls with impunity, and the four ways I was wrong about it
Across eleven European leagues (58,013 completed matches, 115,984 team-matches, and 246 clubs with 150+ matches), raw yellow-cards-per-foul ratios make almost every dominant domestic club in Europe appear leniently officiated (r = −0.371). That apparent impunity is a compound artifact of three confounders: pre-match win expectation (1.098 heavy underdog vs. 0.856 heavy favourite median across all eleven leagues), non-foul conduct cards (dissent, delaying restarts, simulation) entering as a positive affine intercept (11.1%–38.8% league-mean share; 13%–47% across league-seasons), and fixture-level competitive intensity shared by both teams on the pitch (r = +0.283 on opponents' booking indices; 11 of 11 leagues rising across opponent-strength terciles 0.938 to 0.993 to 1.088).
- StudyWhat free football data can still tell you
Public European football repositories exhibit severe longitudinal truncation, unrecorded rule differences in second-yellow accounting, non-foul conduct cards in the yellow-card numerator, and missing referee identifiers across 9 of 11 European leagues. Across 286 completed league-season files in 11 top divisions (2000-01 to 2025-26), only 179 files (62.6%) carry foul and card statistics (58,013 completed matches), while the remaining 107 files return HTTP 200 and parse cleanly into tables containing zero match events. Referee identity is recorded in only 55 of 286 files — continuously in England and Scotland, for two early seasons in Germany and Italy, and never in the other seven leagues.
- TalkHagen: How our AI support agent works
YugabyteDB Friday Tech Talks, Episode 162
How a production AI agent drafts the first root-cause analysis for a distributed database. A single customer diagnostic bundle can hold 50 to 100 million lines of logs. A deterministic layer extracts and normalises the evidence first; the model reasons over what that layer prepared rather than choosing what it reads. Covers the architecture, what worked, and where the system still falls short of its own target.
- Peer-reviewedFrom TextBlob to LLM Agents: Sentiment Model Selection for B2B Technical Support with CSAT Ground Truth
ACL 2026 — Industry Track
A five-year case study of sentiment model selection for customer satisfaction (CSAT) prediction in B2B technical support. The evaluation uses the complete population of CSAT-rated tickets from an enterprise software company: over 500 tickets comprising ~2,500 customer comments from 100+ organizations over five years. 17 approaches across 5 paradigms, plus 11 fine-tuning experiments.
Earlier work, elsewhere
The subject changed four times before this site: SAP, distributed databases, running teams, applied AI. This is the record of that, from conference talks, an employer blog and a personal archive. Not reviewed, and not held to the standard above. This site is the first time the method was written down.
- Pythian blogLeading Through Crisis – What Ernest Shackleton Can Teach Us About the COVID Pandemic
In January 1915, polar explorer Ernest Shackleton's ship became trapped in ice near Antarctica. For the next two years, he kept his crew of 27 men alive on a drifting ice cap, before leading them to safety.
- Pythian blogCassandra Vulnerability - CVE-2020-13946 - Apache Cassandra RMI Rebind Vulnerability
On September 1, 2020, Apache disclosed a security vulnerability for Apache Cassandra. It's possible for a local attacker without access to the Apache Cassandra process or configuration files, to manipulate the RMI registry to perform a man-in-the-middle attack and capture user names and passwords used to access the JMX interface.
Apache Cassandra · JMX
- TalkCapacity Planning Using Facebook Prophet
DataStax Accelerate 2019, National Harbor, Washington DC · with Valerie Parham-Thompson · recording, 26 min
“When will my server run out of resources?” This is a question on the minds of many experienced database administrators. This is particularly relevant in Cassandra given that, as data grows, compactions get triggered to keep the database healthy.
Facebook Prophet · Apache Cassandra
- TalkDocumentation, documentation, documentation!!!
Distributed Data Summit 2018, San Francisco · no recording
The purpose of this talk is to discuss the importance of the documentation in an open source project such as Cassandra, and showing how easy it can be for anyone to build and improve the Apache Cassandra documentation.
Apache Cassandra
- TalkMulti-Cloud Cassandra with Terraform and Ansible
DataOps Barcelona 2018, Barcelona · with Carlos Rolo · recording, 30 min
Given Cassandra nature to scale, automation is an important tool to know. Also, to improve resilience, some go to multi-cloud deployments.
Apache Cassandra · Terraform · Ansible
- TalkCassandra B side, the errors we made and what we learned
DataOps Barcelona 2018, Barcelona · with Carlos Rolo · recording, 42 min
We are with Cassandra since 0.6 and while normally we like to talk about our feats, our experience also comes from our mistakes. In this talk, we will go over some of our (big) mistakes and lessons learned that way.
Apache Cassandra
- Pythian blogBackup strategies in Cassandra
Cassandra is a distributed, decentralized, fault-tolerant system. Data is replicated throughout multiple nodes (centers) across various data centers.
Apache Cassandra
- TalkBuilding an Enterprise Grade App Using Apache Cordova and OpenUI5
ApacheCon North America 2017, Miami · no recording
There is a need growing need for fast deployment of enterprise-grade mobile applications and web apps responsive to all devices and running on the browser of your choice.
Apache Cordova · OpenUI5
- TalkIoT for a Fleet: Building a Route Optimization Platform Using an Open Source Stack
ApacheCon North America 2017, Miami · no recording
Fleet ownership and maintenance can be a big part of a company's cost. Pedro will tell you how to create a competitive advantage by reducing operating expenses and increasing profits with telematics and route optimization, using an Open Source Stack including Cassandra DB, OpenStreetMap and Optaplanner.
Apache Cassandra · OpenStreetMap · OptaPlanner
- 2010 to 2016ArchiveABAP Development Blog
32 posts on Blogger, the longer SAP record: business rules, Web Dynpro, unit tests, SAP from Python.
ABAP · SAP · BRFplus · Web Dynpro · ABAP Unit · PyRFC
- 2013 to 2015ArchiveSAP and ABAP notes
10 notes on a WordPress blog that once answered on this domain. Its dated paths still arrive here and are forwarded there.
ABAP · BRFplus · SAP