Skip to content
Aditya Karnam
AI researcher building the infrastructure layer for reliable agents.
Resume

local-llms

View all tags
Distilling Ornith-1.5-9B: a 4.9 GB Model That Beats the 8.9 GB One It Replaces

— ai, local-llms, open-source

LoRA Fine-Tuning Crashed on Three MoE Models in mlx-lm. The Fix Was One Line, Three Times.

— ai, local-llms, open-source

A Free Local 9B Model Optimized My Rust Agent Harness. Two of Its Three Claims Were Wrong.

— ai, agents, open-source, local-llms

Speculative Decoding on Apple Silicon: 119% Faster on Qwen3.8, Until It Isn't

— ai, local-llms, open-source

The Batch Size That Breaks Local LLM Serving on Apple Silicon

— ai, local-llms, open-source

Why Batching Doesn't Fix Decode: GEMV vs GEMM on Apple Silicon

— ai, local-llms, open-source

MLX Errors on Apple Silicon: A Reference

— ai, local-llms, open-source

Why MLX Decode Slows at Long Context

— ai, local-llms, open-source

Ollama vs vLLM vs SGLang on Apple Silicon

— ai, local-llms, evals, open-source

subagent-fleet: Local AI Compute Control Plane for Coding Agents

— ai, agents, open-source, local-llms

© 2026 Aditya Karnam. AI Researcher.
NowStackField NotesCurrent SystemsStatus