Recent content on Rishi Kulkarni | Statistics & Machine Learning
Wen et al. show two forecasters with identical marginal predictions can make arbitrarily different decisions. A disjoint contextual bandit (one model per arm) is exactly their independence forecaster. In conjugate linear models the joint is exact and free, and partial pooling through a shared component gets it back.
Our most-hit endpoint took 200 milliseconds per shift, so someone capped it at one page of results. The cap ran for years and quietly made product decisions nobody agreed to: an eleven-item shelf, a two-second page, and features that never became ideas. Then a blocked MySQL upgrade forced us to finally read the function.
If you’re running a Bayesian model in a non-stationary environment, you need to forget old data. The obvious approach – scale the precision matrix by a constant – has a failure mode called covariance windup. This post works through three forgetting rules, ending with one borrowed from adaptive control that dominates the others.
My wife has a blog that I self-host. WordPress-based, so it’s MySQL, PHP-FPM, nginx. I’d like to think about its deployment/infrastructure approximately never.
I probably overuse the normal-inverse-gamma posterior. Every time I build a bandit system, every time I need uncertainty quantification for sequential decisions, I end up back at conjugate linear regression.
For a new project at work, we didn’t want to do passwords. The two candidates were magic links and WebAuthn, and we ended up going with magic links—but I got curious about WebAuthn anyway, so I built a demo app to understand it.
Suppose you’re choosing a continuous value x and observing a noisy reward y. The reward depends on x through some unknown function f(x), and you’re making decisions repeatedly—learning as you go. This post explores how to build scalable Bayesian models for this problem using principled approximations.
In my previous post , I covered running portable services—version-controlled config, atomic updates, zero-downtime restarts, all without a container runtime. This is only half of the container story, though. You also need to build the service images.
I run bowl.science , an online Science Bowl tournament platform. It’s a side project, but it’s real production: the DOE Office of Science uses it for their Science Bowl competitions. When a tournament is happening, the app needs to work. There’s no “we’ll fix it in the next sprint.”
My friend Alexa is a graphic designer. Last week I asked her what percentage of her time she spends on meta-design. She asked me what that meant.
Everyone Wants a Churn Model Rarely do I ever get asked to make churn estimates for someone who needs to bring the full power of a proportional hazards model to bear. Besides, the person asking for churn estimates doesn’t actually want to know “what is the probability someone churns eventually?” (Spoiler: it’s 1.)
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. Our question: Does LPS activation significantly increase PNA signal?
Microglia secrete the membrane-tethered glycolipid sialidase, neuraminidase-3, associated with extracellular vesicles affecting the disconnection of neuronal networks. This novel mechanism provides insight into how neuroinflammation disrupts brain function at the network level through contact-independent glycocalyx remodeling.
A while back, I wrote a short piece about planning scientific projects for New Science. The article explores systematic approaches to identifying impactful research questions and structuring PhD projects for maximum scientific contribution.
Hierarchical resampling is a powerful statistical method for analyzing arbitrarily nested experimental designs. This approach combines bootstrap resampling and permutation to control Type I error rates while preserving utilizing all available information.
Fast changes in membrane potential drive neuronal physiology, yet observing neuronal activity noninvasively remains challenging. We report RhodolVoltageFluor-5 (RVF5), an optical voltage reporter with improved photostability and brightness under both one- and two-photon illumination. RVF5 enables robust voltage imaging in thick tissue and brain slices, revealing increased neuronal activity in a…
Rishikesh Kulkarni Vice President, Engineering and Machine Learning/Data Science 📍 Lexington, MA | 📧 rishi@kulkarni.science | 🌐 rukulkarni.com | 💻 github.com/rishi-kulkarni | 💼 linkedin.com/in/rishi-kulkarni