🇰🇪📍 kenya-locations v0.11 is out
every county,constituency, ward,sub-county,locality + area in Kenya ~4,838 places as one open dataset.
(so far) but welcome to add more
search any place and use it in JS,React,Kotlin,Swift, Dart,PHP or WordPress.
→ https://t.co/d60jqjaZSO
A candidate mentioned building RAG applications on his resume, but one interview question caught him off guard.
The interviewer asked:
“Your RAG system works fine with small documents, but when users start uploading hundreds of pages of PDFs, the answers become less accurate. How would you fix that?”
He suggested increasing the Top-K value to retrieve more chunks.
The interviewer asked:
“But won't that also bring in more irrelevant information and confuse the LLM? What else would you try?”
He couldn't give a convincing answer.
What would you have done differently?