work

i am in the business of prophecies. the world, much like jean land, runs on a multiplicity of tomorrows—predicted, imagined, predestined visions of how things would go, how things could go, how things ought to go, you get the idea. to be good at this work you need to know a lot about the world: and so it goes that some worldly people are inadvertently prophets.

more concretely, predicting the future is a game where, given everything you know about the past, you are deciding what to do today because of what you think will happen tomorrow. once you have system of generating prophecies for the future, you want to consider the following three things:

1. how this same prophecy played out in the past
2. what to do, given your belief in your prophecy
3. what's the worst that can happen, if the prophecy fails.

usually looking at 1 gives you a better idea of 2, and you work with a list of constraints that helps you think about how to optimize 2 given your tolerance of 3. my work consists of writing python code to answer the above questions, and repeatedly banging my head against the wall. on some days i get asked to do data analysis (usually with spark, pandas, sql) or write some small ai python applications (basically 17 llm api calls in a trenchcoat, usually with Great Value homemade rag for guerilla document processing and/or dspy)