Not one of my typical posts, but I'm excited to share that I've joined AfterQuery as a Strategic Projects Lead.

I believe that there will continue to be a need for high-quality training data for the world's leading AI models in the coming years. While models can now effectively reason through problem sets, there is still a dearth of data that focuses on codifying real-world expertise in ways that can push model performance meaningfully forward. A real step-change in productivity can come from training data that better reflects how experts actually work in real workflows.

Those who know me know that I've always been interested in people and data — what can data reveal about human behavior, incentives, and decision-making. In college, I was drawn to social networks as models of quantified, stored value on edges and nodes. I was fortunate enough to have the opportunity to bore professors at Duke with outlandish ideas related to networks and people, from modeling online radicalization dynamics to testing how much hype affected NFT art prices (I'll spare the details of the more ill-advised research ideas).

By senior year, I started reading LessWrong and trying (unsuccessfully) to @Eliezer Yudkowsky on X because I became interested in LLMs and thought experiments around mis-specified optimization problems (including non-human-oriented outcomes from AI systems generally).

Now, a few years later, I've come around to the idea that the next evolution of model capability will involve training models on how people actually reason through competing priorities and, more importantly, how they respond through messy human problems people experience as they work.

And that's precisely what AfterQuery is focused on. By building datasets for our customers that translate expert domain knowledge (across finance, law, medicine, and more) into training data that improves model performance and efficiency, we help models improve in answering the kinds of questions humans work on every day, along with the kinds of questions that are on the horizon.

This role comes with a personal update. I've left consulting and moved from New York to San Francisco (my second Bay Area stint in two years). Leaving the city behind was hard because it began to feel like home. Fortunately, my incredible friends and family were understanding. After all, this career move came with unusually high conviction. So far, the work has been meaningful: a steep learning curve and real responsibility. I'm grateful for the opportunity and I'm excited to get to work.

AfterQuery is growing quickly and we are hiring across growth, engineering, and operations. If you're interested in working here or in becoming an expert with us, please reach out on LinkedIn or at my new email: kishan@afterquery.com.