16 - Richard Eisenberg on OCaml, Effective AI, Teaching FP and Hiring for Fundamentals

Richard Eisenberg is a language designer and compiler engineer for OCaml at Jane Street and a core contributor to the Haskell language. He focuses on static type systems and functional programming to make software more reliable, while maintaining ease of use and runtime efficiency. Richard has also taught various computer science topics to both university and high school students, and he holds a PhD from the University of Pennsylvania.

You can find Richard's work at https://richarde.dev/

0:00 Intro
0:56 Teaching CS to high school students
3:20 Turning down Microsoft for a teaching career
8:26 PhD, faculty life, and joining Jane Street
12:32 Algorithms and FP for high schoolers in the age of AI
17:08 How Richard uses LLMs day-to-day
22:29 Jane Street's internal AI experimentation
23:08 OCaml in the age of AI: open sourcing and error messages
29:31 Making illegal states unrepresentable: how far is too far
38:11 Formal verification and LLMs
42:08 Effect handlers in OCaml
44:03 Rust-style ownership and data race freedom in OxCaml
51:06 Why Jane Street chose OCaml
57:54 OCaml vs Haskell: purity, mutability, and physical equality
58:47 Onboarding engineers to OCaml at Jane Street
1:00:01 Hiring for fundamentals, not OCaml experience
1:07:07 Screening for good programming taste
1:10:47 Training vs hiring: the chicken-and-egg problem
1:15:31 Why OCaml hasn't gone mainstream
1:25:08 Dynamic languages: Python, JavaScript, and their place
1:28:15 Rust, C++, Scala: languages Richard finds interesting
1:33:08 Research at Jane Street: papers, peer review, and process
1:36:17 Contributing back to open source ecosystems
1:37:46 Dependent types in Haskell: motivation and the long road