Uploaded March 2024 | Updated September 2026, 3 weeks ago
Check out my essays: aisc.substack.com
OR book me to talk: calendly.com/amirfzpr
OR subscribe to our event calendar: https://lu.ma/aisc-llm-school
AF: LLMs are not sitting in the vacuum. They're interfacing with your databases. They're interfacing your retrievals, probably some biological models, physical models. How do you go about designing a system around LLMs to make sure it is robust, make sure it delivers, reduces the hallucination, has verifications, to make sure that is actually reliable, and all of the other nuances that might be relevant in your domain.
AM: That's a great question! When we think about those, we have to remember in applications like drug discovery elements are not alone. It's not like we push a button and then say, okay, just spit out the compound and go for clinical trials.
We have experts that sit on top of different stages of this process. Those experts are not there only to push the programs forward, but also they can provide feedback for improving, for monitoring our models for the LLM itself. At the end of the day, if you're dealing with LLM in expert domain tasks, like drug discovery, we need those. So we need to have biologists, medicinal chemists, computational chemist people, computational biology people.
Check out my essays: aisc.substack.com
OR book me to talk: calendly.com/amirfzpr
OR subscribe to our event calendar: https://lu.ma/aisc-llm-school
AF: LLMs are not sitting in the vacuum. They're interfacing with your databases. They're interfacing your retrievals, probably some biological models, physical models. How do you go about designing a system around LLMs to make sure it is robust, make sure it delivers, reduces the hallucination, has verifications, to make sure that is actually reliable, and all of the other nuances that might be relevant in your domain.
AM: That's a great question! When we think about those, we have to remember in applications like drug discovery elements are not alone. It's not like we push a button and then say, okay, just spit out the compound and go for clinical trials.
We have experts that sit on top of different stages of this process. Those experts are not there only to push the programs forward, but also they can provide feedback for improving, for monitoring our models for the LLM itself. At the end of the day, if you're dealing with LLM in expert domain tasks, like drug discovery, we need those. So we need to have biologists, medicinal chemists, computational chemist people, computational biology people.
