Hey everyone! I've been mulling a fun idea around ...
# ml-and-mlops-questions
p
Hey everyone! I've been mulling a fun idea around for awhile that I want to try. Every week I want to try picking a practical AI topic to spend at least ~1 hour learning. I'll start a thread each week for async communication and project/resource sharing. At the end of the week I'd like to try a casual live stream for showing what we've built + discussions. Similar to #C06AAG8LU9M I'll be using Flyte to build my pipelines, but you're not required to if you want to participate! So far this is my list, would love to know what else you'd like to add! • MCP Servers • Agentic search with Tavily • Reinforcement learning with OpenENv
🔥 2
👍 1
❤️ 1
c
This sounds awesome. One area I’ve been wanting to dig into is prompt evaluation for AI apps—like how to systematically test and measure prompt quality over time instead of just eyeballing results.
🙌 1
p
I'll add it to the list!
👍 1
a
@clever-businessperson-20427 You should look into BAML https://boundaryml.com
🙌 1
💯 1
p
BAML is really cool! Lets definitely do a week on that. I know the founders a bit, maybe I'll see if they want to hop on the stream or do something around it!
a
There is a prompt optimization technique that we use called “GEPA”. We have our own version of it, but BAML also provides a OOB version of it. https://docs.boundaryml.com/guide/baml-advanced/prompt-optimization Paper: https://arxiv.org/pdf/2507.19457
🔥 1