Skip to content
Home

Blog Posts

Research perspectives on reliable AI, data systems, and optimization.

Diagram showing an LLM occupying most of a laptop’s 128 GB memory, leaving limited room for context and inference.

I Tried Running Local LLMs on a Maxed-Out MacBook. Here’s What Happened.

How useful are local LLMs on the most powerful laptop you can buy? I tested several leading open-weight models on a real coding task, then repeated the ...

Keywords: Local LLMs · Open-weight models · LLM benchmarking · Coding agents · MacBook Pro · Apple Silicon · NVIDIA H100 · AI model evaluation · Context window · Model quantization

Read article
How DAGSmith optimizes a dbt pipeline

Why Slower SQL Can Make Your dbt Pipeline Faster

Modern analytics is increasingly organized as recurring SQL pipelines rather than isolated SQL statements. Tools like dbt have gained extreme popularity in ...

Keywords: dbt · query rewriting · database optimization · snowflake · databricks

Read article