Why one researcher's immunology model was the code he'd only ever write by hand
iskander · x · 2026-11-20
In this thread segment, researcher @iskander describes why his joint MHC-I/II model (with expression-parameterized antigen processing and CD4/8 immunogenicity outputs) was so hard: "conceptually really complicated, easy to break one part working on another." Training and eval are slow, the model is intricately interwoven with how he curates niche immunology datasets, and it needs many domain priors injected.
This is why progress was slow — it was the one piece of software he'd have written by hand.
Related event: Researcher pits multiple AI models on immunology code with striking gaps(4 posts)→
More from coding & agent
- Where does the agent end and plain code begin? One dev's rule for splitting workflows — krunal_builds · 2026-11-20
- Grok Bot 1.2.3 ships flight booking, per-tool permissions and community directory — NicoVerderosa · 2026-11-20
- A 2026 company is just a folder: every department as .md agent files — mdancho84 · 2026-11-20
- Banks warn customers they could be liable for their AI agents' mistakes — beingmodest · 2026-11-20
- Google AI Studio can now build Android apps from a prompt and install them privately, no Play Store listing needed — TawohAwa · 2026-11-20
- No task tracker fits agentic dev: 10-100 concurrent tasks need new patterns — bigblueboo · 2026-11-20