5 min read
Building synthetic data pipelines: Ray
Using Ractor vs Ray for building synthetic data pipelines
Pradeep - notes on AI/ML, system design, and building software
Featured
Understanding LLM Token Caching
The missing details you need to run production prompt caches: provider minimums, write costs, TTL/eviction, and how the KV cache actually maps to your bill.
12 min read
Latest
All posts- 7 min read
Building synthetic data pipelines: Ractor
Using Ractor vs Ray for building synthetic data pipelines
- 17 min read
Deploying and Operating RAG Systems
Enterprise patterns for building reliable, scalable RAG systems with clear service boundaries and governance layers.
- 17 min read
Microservices Architecture - Building Scalable RAG Systems
Enterprise patterns for building reliable, scalable RAG systems with clear service boundaries and governance layers.
- 14 min read
The Multi-Hop Retrieval Pipeline - From Query to Answer
Enterprise patterns for building reliable, scalable RAG systems with clear service boundaries and governance layers.
- 9 min read
Why RAG Needs Multiple Retrieval Layers
Enterprise patterns for building reliable, scalable RAG systems with clear service boundaries and governance layers.
Projects
All projectsAdTurbine
AdTurbine is a platform for generating UGC (User Generated Content) ad videos using AI. The system follows a pipeline architecture.
fastapinext.jsdockerllm
Ragforge
Enterprise patterns for building reliable, scalable RAG systems with clear service boundaries and governance layers.
fastapipgvectorprometheusnext.jsdocker