o latent Family Source

Skip to content

kernel ready 3 cells

cells

Evaluating LLM outputs without fooling yourself
MCP: giving language models real tools
RAG pipelines that actually work in production

all 3 cells →

namespaces

Evals 1
Retrieval 1
Systems 1

navigate

series
tags
archive
about

Index

Categories

3 categories across 3 posts, filed under 2 desks.

Systems

Systems 1

Evals

Evals 1
Retrieval 1

run complete — ship it, then log what broke.

categories/

all
Evals
Retrieval
Systems

tags/

#Agents
#Benchmarks
#Embeddings
#Evals
#LLM
#LLM-as-Judge
#MCP
#Metrics

nav/

series
archive
now
about

Legal, family, and brand

Legal

Privacy
Terms
Refunds
Disclaimer

Family

About
Contact
Sitemap
Security.txt

Connect

Newsletter
RSS
GitHub
Sponsors

A working log on building AI systems that survive production — RAG, agents, and honest evals — by Chirag Singhal.

© 2026 Chirag Singhal · latent · MIT

home Home read Latest series Series find Search menu Menu