hi! i’m neal

i’m making machines more reliable. i’m also learning ml security so i can help solve the world’s toughest problems.

right now i’m a formalization engineer at millennium research (contract), where i’m evaluating ai coding agents. i spend a lot of time somewhere between mathematics, programming, and machine learning.

lately, i’ve been poking at how we evaluate math models and working on proofs in lean. previously, i did machine learning research at jhu apl through the aspire program.

outside of that, i’m usually following a question much further than i meant to.

here are some things i believe in:

curiosity doesn’t need a business model
make the thing. then make it better.
taste starts with noticing
take ideas seriously. yourself, slightly less.
good conversations expand what feels possible
leave some room for side quests

here are some things i like:

cult member, songs that get better after the tenth listen, bass you can feel in your chest, seamless album transitions, attack on titan foreshadowing, refusing to skip a good opening, mangoes eaten over the sink, window seats, rain when i don’t have anywhere to be, the cold side of the pillow, warm lamps, sending someone a song with no explanation, laughing so hard the story never gets finished, friends who stay for “one more song”, blackboards, hagoromo chalk, excalidraw.

things i don’t like:

spoilers disguised as hints, mangoes that need another three business days, wet socks, volume jumps between songs, someone talking through the part i wanted them to hear.

i’m collecting notes and things i figure out on the blog.

good questions and unexpected conversations are welcome. send me a message! (email works best)