Rishi Kulkarni | Statistics & Machine Learning

A little performance vignette.

My old churn model measured how surprising a user’s silence was against their own history.

Wen et al. show two forecasters with identical marginal predictions can make arbitrarily different decisions.

Our most-hit endpoint took 200 milliseconds per shift, so someone capped it at one page of results.

If you’re running a Bayesian model in a non-stationary environment, you need to forget old data.

My wife has a blog that I self-host.

I probably overuse the normal-inverse-gamma posterior.

For a new project at work, we didn’t want to do passwords.

Suppose you’re choosing a continuous value x and observing a noisy reward y.

In my previous post, I covered running portable services—version-controlled config, atomic updates, zero-downtime restarts, all without a container runtime.

I run bowl.science, an online Science Bowl tournament platform.

My friend Alexa is a graphic designer.

Update, August 2026: I no longer stand behind this approach.

A Motivating Example We were studying how microglia affect neuronal networks using a standard imaging experiment: 3 mice, 3 coverslips per condition, about 20 neurons measured per coverslip.

Microglia secrete the membrane-tethered glycolipid sialidase, neuraminidase-3, associated with extracellular vesicles affecting the disconnection of neuronal networks.

A while back, I wrote a short piece about planning scientific projects for New Science.

Hierarchical resampling is a powerful statistical method for analyzing arbitrarily nested experimental designs.

Fast changes in membrane potential drive neuronal physiology, yet observing neuronal activity noninvasively remains challenging.