Empowering Developers with local AI
summary
Claims that full local workflows replace cloud Copilot and ChatGPT with open-source models and tools and demonstrates it working: running Quen 2.5 coder 7B, DeepSeek R1 7B and Llama 3.2 3B locally, swapping GitHub Copilot for continue.dev and chat for Open Web UI, exposing a REST API and codebase indexing via LM Studio and Ollama. Reports performance: DeepSeek ~50 tokens/s, Llama ~28 tokens/s on a GPU, shows code edits succeed but also flaky answers (miscounts "strawberry"), and verifies offline operation and indexing actually reduce latency.
AI-written summary · about
chapters (10)
- 0:00Local AI for developers
- 1:42Introduction
- 2:47AI usage trends
- 4:57Why local AI matters
- 6:43Running LLMs with LM Studio
- 10:23Local model management with Ollama
- 14:46Replace GitHub Copilot with Continue.dev
- 22:04ChatGPT alternative with Open Web UI
- 31:54Performance benchmarks & hardware tips
- 35:33Call to action for local AI
related talks
48mMastering the AI Shift: Adaptive Strategies for Software Evolution
46mThe AI-Powered Developer
27mAI Demystified Unveiling the Intelligence Behind the Machine
32mBeyond the Hype: A Pragmatic Look at How Developers Actually Use AI Tools
05mBuilding Your (Local) LLM Second Brain
55mBuild enterprise generative AI apps using Llama 3 at 1,000 tokens/s on the SambaNova AI platform
45mRunning Your Coding Agent Locally: Lessons from a Real-World Experiment
29mGenAI in Backstage - Turbocharge Your Developer Productivity!