Data For Science wire the eval suite into a CLI command and hook it into GitHub Actions to create a CI gate.
Public source
Publisher name
Public post
People ask me: "Can you really build a production eval harness in four hours?" Yes. Here's exactly what we build. Hour 1 — The skeleton. We set up Inspect-AI, define the…
Company
Data For Science
Take Control of Your Data
- Industry
- Technology, Information and Internet
- Location
- New York, US
- Company size
- 1 employee
About Data For Science
Data For Science helps organizations deploy AI systems that are reliable, measurable, and maintainable. We combine deep technical work with clear communication and practical training.
See moreLatest activity
Latest activity from Data For Science
6 signals
Research & Knowledge
Data For Science published an essay exploring how Large Language Models are the new Wikipedia.
Research & Knowledge
Data For Science published an essay exploring how Large Language Models face skepticism about reliability and authority similar to Wikipedia in the early 2000s.
Products & Services
Data For Science builds a production eval harness in four hours with Hour 1 running the skeleton, Inspect-AI, dataset loader, and first eval run passing.
Discover more
Similar signals
Similar public activity from other companies.
Products & Services
Lander Analytics
Lander Analytics builds standardized development platforms including CI/CD workflows, automated testing, documentation, security scanning, dependency tracking, agent skills, and properly configured development environments.
Products & Services
GitHub
GitHub extended HyDRA from model selection to workflow orchestration
Products & Services
Dailybot
Dailybot introduced a new system on the CLI and API side to push activity tied to a person with a message and optional link, enabling activity tracking on the next check-in report.
Products & Services
WorkOS
WorkOS Blog Bot ships a verified fix in a PR in three minutes from a bug report, as the triaging person already knows which reports matter.
Products & Services
Workloft.ai