Let's Data Science

@letsdatascience

Expore . Learn . Achieve

New Delhi, India

Joined July 2023

2 Following

52 Followers

Replies Reposts Media

Let's Data Science
@letsdatascience
Nov 4
Vector search put a retired rule first. The current one came ninth. Our free tutorial with @ActianCorp shows how to build an on-prem AI agent that gets this right: filtered search, a local model and human-approved memory. Full code on GitHub: letsdatascience.com/blog/build-on-…
Let's Data Science
@letsdatascience
Oct 30
Your vector database runs inside your network. Your data might still leave it. "Keeping the database local doesn't automatically mean the data remains local." @ActianCorp CTO @emmakmcgrattan tells LDS where local AI can still expose data. letsdatascience.com/news/actian-ct…
1
Let's Data Science
@letsdatascience
Oct 27
An AI agent arrives with a verified signature. Can it buy? A signature "only says which product is calling," @ArkoseLabs founder and CEO @kgosschalk tells LDS. His rule: "welcome agents that sign but judge every session on behavior, signed or not." letsdatascience.com/news/arkose-la…
1
Let's Data Science
@letsdatascience
Oct 27
RAG apps may run several models per document: one rewrites the query, one answers, one checks. @IBMResearch's Luis Lastras tells LDS when Granite Switch should do it all in one model. On IBM's RAG demo: "the single model solution is far more efficient." letsdatascience.com/news/ibm-tells…
1
Let's Data Science
@letsdatascience
Oct 27
An AI agent sends a purchase order. The ERP times out. Retry, and you might order twice. "When the ERP times out, we don't assume the write failed." @Aera_Technology co-founder and CTO Shariq Mansoor tells LDS what an agent must check first. letsdatascience.com/news/aera-tell…
1

Let's Data Science

Let's Data Science’s posts