How a Christmas-1989 hobby project descended from a teaching language became the default tongue of scientific computing and machine learning — through a decade-long, self-inflicted version migration, a global interpreter lock it still fights, and a benevolent dictatorship that resigned over a walrus.
Machine-Learning
-
The Story of Python -
Embedding Models and Vector Search, Honestly A multi-line description of two to three lines explaining what the post covers: what an embedding represents, dense vs sparse vs hybrid retrieval, HNSW and IVF-PQ ANN indexing, and where ranking quality actually lives.
-
RAG Beyond Toy Demos: Chunking, Reranking, and the Evaluation Problem Nobody Solved The gap between a notebook RAG demo and a system that ships: why chunking is the retrieval you never tuned, query rewriting and HyDE, the retrieve-wide rank-narrow pipeline, citation faithfulness and grounding, and the evaluation problem that quietly decides whether any of it works.
-
Training Cluster Networking: NVLink, Rails, and the Bandwidth Budget That Bounds Model Scale Why the network, not the GPU, decides how big a model you can train: the NVLink and NVSwitch scale-up domain versus the InfiniBand and RoCE scale-out fabric, rail-optimized topology, in-network reduction, and the bandwidth-budget math that maps tensor, pipeline, data, and expert parallelism onto real cables.
-
Autofocus Systems Explained Autofocus is the camera feature most photographers stopped thinking about decades ago, and it has been quietly reinvented twice since then. We walk contrast-detect versus phase-detect history, the on-sensor phase-detect revolution that mirrorless made standard, the neural-network subject and eye detection that defines 2026 flagship performance, and the honest gap between marketing claims and what actually focuses in real shoots.
-
Computational Photography A modern phone camera captures one image to display and computes perhaps a dozen behind it, fusing them into a result no single exposure could deliver. We walk what the phone's pipeline actually does between shutter press and saved JPEG: multi-frame alignment, HDR fusion, night-mode stacking, semantic segmentation for portrait mode, and the honest line between optical capture and after-the-fact reconstruction.
-
Shannon and Information Theory: The 1948 Paper That Named the Bit Claude Shannon's 1948 paper defined the mathematical foundation of every digital communication system on earth. This post unpacks entropy as surprise, the source and channel coding theorems, the Shannon-Hartley limit, and the unexpected appearances of Shannon entropy in machine learning and the Kelly criterion.
-
Ray: Distributed Python Without the Pain A deep technical look at Ray's core primitives, scheduling model, high-level ML libraries, KubeRay on Kubernetes, and honest trade-offs versus Dask and Spark.
-
Local LLMs with Ollama Running large language models locally with Ollama — VRAM requirements by tier, quantization formats explained, model selection for different use cases, Open WebUI setup, modelfiles for custom system prompts, and an honest comparison of when local beats cloud and when it doesn't.
-
Immich: Self-Hosted Google Photos Deploying and running Immich as a self-hosted photo and video library: Docker Compose setup, PostgreSQL with pgvecto.rs, machine learning face recognition and CLIP semantic search, hardware transcoding, external library mounting, and a solid backup strategy.
-
Edge AI Accelerators: Coral, Hailo, and Jetson Orin Nano Super Compared A practical guide to edge AI accelerators for the homelab: Google Coral TPU, Hailo-8 and Hailo-8L, and the Jetson Orin Nano Super. Real benchmarks, power draw, use case fit, and where each beats or loses to a small discrete GPU.
-
Mixture of Experts Internals: Routing, Expert Parallelism, and Load Balancing MoE is now the default architecture for top-tier open models. This is how it actually works: router design, top-k token dispatch, auxiliary loss for load balancing, expert parallelism across GPUs, and what it means for inference serving — with DeepSeek-V3 and Mixtral as concrete case studies.
-
FSDP and DDP: Distributed Training Patterns That Actually Scale The distributed-training patterns that actually scale in PyTorch — DDP versus FSDP, how each shards parameters, gradients, and optimizer state, and the collective-communication and precision knobs that decide whether more GPUs help.
-
NCCL Deep Dive: Multi-GPU Collectives, Ring vs Tree, and Debugging Distributed Training How NVIDIA's NCCL moves gradients across multi-GPU and multi-node training — ring versus tree collectives, NVLink and InfiniBand topology awareness, and how to debug the silent hangs and bandwidth cliffs that take training down.
-
Google's TurboQuant: What It Means for Home AI Enthusiasts Google Research's TurboQuant compresses LLM KV cache memory by 4–5x with minimal accuracy loss. Two weeks in: real benchmarks, active framework integrations, controversies, and what you can actually run today.
-
Immich Deep Dive: AI-Powered Photo Management for Your Homelab A complete guide to deploying and running Immich — self-hosted Google Photos with AI-powered face recognition, smart search, and hardware-accelerated machine learning.
-
MLOps Fundamentals: Experiment Tracking, Model Registry, Serving, and Drift Monitoring A practical guide to MLOps — structuring experiments with MLflow, managing the model lifecycle through a registry, serving models in production with BentoML and Triton Inference Server, and detecting data and concept drift before it silently degrades your models.
-
Fine-Tuning Small Models: When and Why to Fine-Tune vs. Prompt Engineer A practical guide to fine-tuning small language models — understanding when fine-tuning beats prompt engineering, how LoRA and QLoRA work, training a model on your own data with Unsloth or Axolotl, and deploying the result with Ollama.
-
Local LLM Deep Dive: Ollama, Quantization, and Running AI on Your Own Hardware A comprehensive technical guide to running large language models on your own hardware — covering Ollama setup, quantization formats, hardware selection, Apple Silicon and NVIDIA GPU configuration, and building a private local coding assistant.