AI Engineering
Building with AI models: training, evaluation, retrieval, serving and voice.
Fine-tuning Whisper, Parakeet and other speech models on your own voice and languages, and measuring the result.
Benchmarks 22 episodesPublished AI benchmarks, what they measure, and where their methodology breaks down.
Evals 33 episodesBuilding and running your own evaluations: harnesses, LLM-as-judge, statistics and model bake-offs.
Fine-Tuning 38 episodesAdapting models after pretraining: LoRA, SFT, DPO, RLHF, data size, epochs and cost.
Inference and Serving 26 episodesHow models are served and paid for: batch APIs, inference providers, serverless GPUs, latency and the KV cache.
Local AI and Quantisation 18 episodesRunning models on your own hardware: VRAM budgets, GGUF and quantisation, Ollama and local routing.
Model Context Protocol 23 episodesBuilding, wiring, securing and right-sizing MCP servers and gateways for AI agents.
Prompt Engineering 23 episodesWriting, layering and maintaining system prompts and instructions that models actually follow.
RAG and Retrieval 26 episodesRetrieval-augmented generation, embeddings, vector databases, re-ranking and knowledge graphs for AI memory.
Small Models and Distillation 23 episodesSmall language models, classifiers, distillation and model merging as cheaper alternatives to frontier LLMs.
Synthetic Data 10 episodesGenerating training data with models: privacy, preference data, weak supervision and model collapse.
TTS and Voice Cloning 24 episodesText-to-speech models, Chatterbox, voice cloning, prosody and the pronunciation problems of synthetic speech.
Voice Dictation 27 episodesDaniel's voice-typing stack: mics, triggers, on-device ASR, and cleaning up dictated text.