Evaluating LLM Output Quality: Automated Evaluation with RAGAS and TruLens
Implementing automated evaluation metrics for RAG and LLM applications using the RAG Triad: Faithfulness, Answer Relevance, and Context Precision.
Building Function-Calling Workflows: Tool Use with Anthropic Claude 3 and OpenAI
Software engineering patterns for implementing deterministic function calling and tool execution across Anthropic Claude 3 and OpenAI models.
Autonomous AI Agents: Designing Robust Function Calling and Tool-Use Frameworks
How function calling and structured tool integration allow AI models to perform complex multi-step backend actions.
vLLM and PagedAttention: High-Throughput Serving of LLMs in Production
How vLLM and PagedAttention revolutionize GPU memory utilization, enabling 20x higher request throughput for self-hosted LLM inference clusters.
Reranking Models in RAG: Boosting Context Relevance with Cohere and BGE
Implementing two-stage retrieval in RAG pipelines using cross-encoder rerankers (Cohere Rerank, BAAI BGE) to eliminate retrieval hallucinations.
Small Language Models (SLMs) on the Edge: Deploying Efficient On-Device AI
Why compact Small Language Models (SLMs) are taking over edge computing and local offline AI applications.