Skip to content
Devtools

Local LLM Code Autocomplete

Code completion that runs entirely on your machine.

Advanced22h1–2 people16/20 fit
TL;DR

Build Local LLM Code Autocomplete for developers who cannot send code to the cloud in ~22h with 1–2 people. Stack: Ollama + TypeScript + VS Code API.

Why it wins: Privacy-focused AI coding.

Build flow

01 · Problem
Many companies forbid sending code to cloud AI.
02 · Solution
Serve a small code model with Ollama, build an editor extension for inline suggestions and measure latency and acceptance.
03 · Build
  1. 1Run a model with Ollama
  2. 2Build extension
  3. 3Send context
  4. 4Show suggestions
04 · Demo
Show completions with networking turned off.

MVP checklist

  • Inline completions
  • Local model
  • Latency stats
  • Context selection

Tech stack

TechRole
OllamaLocal LLM runtime
TypeScriptTooling
VS Code APIExternal data

Reference repos

Open-source projects you can build on. Links checked 2026-10-31.

Related ideas

More Devtools