Be part of Fintella Labs.

What we work on.

• Self-recognition is the closest thing to ground truth, and it is not enough.

  In behavioral inference, when a person recognizes the description of themselves, that is face validity, not validation. We work on what stronger validation looks like when no clean external criterion exists.

• Detecting behavioral shifts in real time, with calibrated confidence.

  Phase shifts are easy to spot in retrospect. The harder problem is calibration without labels, on a process that does not sit still.

• The space between what you can claim and what is projection.

  We see choices, not counterfactuals. Drawing causal conclusions from observational behavioral data is a methodological problem that does not get easier when the data is rich.

• Inferring affect from data that was not collected to measure it.

  Whether emotional states can be read off behavioral traces in any reliable way is an open empirical question. We are testing how far this goes.

• What informed consent means when a system tells you something about yourself you did not already know.

  Behavioral inference produces information that did not exist before. The legal and ethical frame for that information is unsettled, and we work in the middle of it.

We'd like to hear from people who work in these areas. Computational behavioral scientists. Psychometricians who think hard about validity. Behavioral economists working with revealed preference. Privacy researchers familiar with reconstruction attacks. ML people who've worked with longitudinal data. Causal inference researchers. And engineers who like building when the abstractions are still moving.

If you'd want to do this with us, write. Tell us who you are, what you've worked on, and why this rather than something else.