Posts RSS feed
2026
-
Learn to fine-tune your own LLMs using Unsloth with LoRA/QLoRA. Covers data preparation, SFT, DPO/ORPO alignment, evaluation metrics, and cost analysis for when self-hosted beats API providers.
-
A comprehensive guide to deploying vLLM for high-throughput LLM inference in production. Covers server configuration, async client patterns, parallel processing, Kubernetes deployment, performance tuning, and monitoring.
-
A head-to-head benchmark comparing pandas, polars, and duckdb. See why modern DataFrame libraries deliver 5-20x speedups through lazy evaluation, predicate pushdown, and automatic parallelization.
2025
-
Building a retrieval-augmented generation pipeline using just the OpenAI API and NumPy—no vector databases, no frameworks, just the fundamentals
-
Build an AI agent with tool use, memory, and error recovery using OpenAI's function calling API. No frameworks—just the patterns that matter.