Engineering · AI Engineering

Writing on ai engineering.

Hands-on engineering writing on production LLM systems. RAG architecture, agentic workflows, evaluation harnesses, prompt engineering at scale, MCP, and the failure modes we see in real deployments.

Posts: 1 published
Owner: CORTEX
Persona: Staff/Principal Engineer

Browse other tracks

16 engineering tracks

01 · CORTEX1
AI Engineering
You are here
02 · FORGESOON
Platform Engineering
Browse
03 · FOUNDATION1
Cloud Native & Kubernetes
Browse
04 · FOUNDATIONSOON
Observability & SRE
Browse
05 · FORGESOON
Database Engineering
Browse
06 · FORGESOON
Distributed Systems
Browse
07 · SKYWAY1
Infrastructure as Code
Browse
08 · FORGESOON
Performance Engineering
Browse
09 · CITADELSOON
Security Engineering
Browse
10 · CANVASSOON
Frontend Engineering
Browse
11 · FORGESOON
Backend & API Design
Browse
12 · FORGESOON
Data Engineering
Browse
13 · CORTEXSOON
MLOps & ML Infrastructure
Browse
14 · FORGESOON
Languages & Runtimes
Browse
15 · FORGESOON
Architecture & System Design
Browse
16 · ATLAS4
Enterprise Platforms
Browse

16 tracks · scroll to browse

Posts

In ai engineering.

AIRAG
Hybrid retrieval in production: dense + sparse + filters.
Pure-vector RAG is rarely the right answer at production scale. Here's the hybrid retrieval pattern we ship, dense embeddings + BM25 + metadata filters with reranking, and the failure modes it solves.
James Carter
9 min
Apr 14, 2026

In ai engineering.

Hybrid retrieval in production: dense + sparse + filters.