Phi9 / research & engineering
Work with Phi9
Bring a concrete question, a system to evaluate or a failure worth understanding.
Research critique
Challenge an assumption, an identifiability claim or an experimental control. A useful first exchange names the result and the test that could overturn it.
Reproducible evaluation
Discuss a bounded comparison with explicit baselines, information access, costs and failure criteria. Proposed output: protocol, executable benchmark and an honest result report.
Representation & execution
Explore task-specific memory, interpretable transformation programs or instrumentation for a research system. Proposed output: a scoped implementation and inspectable traces.
Physical-workflow study
Explore whether an observation study could answer a real operational question. Site permission, participant consent, usage rights and data handling must be agreed before collection.
Start an enquiry
This opens your email application with a short enquiry outline. It does not book an engagement, submit a site form or confirm delivery. Mailbox delivery has not been end-to-end tested as part of this release.
[email protected] is the contact address already configured in the site's source. Do not send confidential datasets, credentials or personal records in an initial enquiry.
What happens next
An initial discussion should settle the question, available evidence, success criteria and permission boundaries. Any scope, fee, schedule or delivery commitment is agreed separately. These are proposed collaboration formats, not claims of existing customers or guaranteed outcomes.
Who is behind the work?
Shubham Attri, founder of Phi9. Read the thesis, inspect the evidence or run the implementation before deciding whether the work fits your question.