Curtis Nix

What I Run, and Why

Sun, 21 Jun 2026 00:00:00 +0000

The reasoning behind the choices in my homelab: why Talos, why Debian as the gateway, why local inference, and where convenience and control trade off.

Debian LUKS Locked Me Out of My Cluster for 14 Hours

Sat, 20 Jun 2026 00:00:00 +0000

An overnight reboot on the LUKS-encrypted Debian server that fronts my Talos cluster halted at the passphrase prompt. The cluster and every hosted service were unreachable until I drove to the box and typed it in.

A Power Outage Cut My Access To My Homelab During Vacation

Sat, 20 Jun 2026 00:00:00 +0000

A power outage during a week-long Charleston vacation took down the Debian server that fronts my Talos cluster. With LUKS at boot and no UPS to ride out the flicker, my remote access to the homelab stayed cut for the whole trip. I came home, ordered an APC Back-UPS 600, and then put off plugging it in. A second outage a few days later was what finally got it installed.

Two-Board ESP32 Lab: ESP-NOW Foundations

Mon, 01 Jan 0001 00:00:00 +0000

Bootstrap of a permanent two-board ESP32 lab with an always-on agent reachable via chat. Validated end-to-end with 1 Hz ESP-NOW pingpong (100/100 frames) and RSSI-based distance estimation.

Custom OTA over ESP-NOW

Mon, 01 Jan 0001 00:00:00 +0000

A custom four-message OTA protocol for pushing firmware between ESP32s over ESP-NOW (no WiFi, no BLE). Protocol and partition layout proven end-to-end on small payloads; the full 742 KB push reaches 97% before crashing on memory pressure.

LLM Backend Eval Round 2: K8s Triage Scenarios

Mon, 01 Jan 0001 00:00:00 +0000

Follow-up eval comparing llama.cpp and MLX on a focused K8s troubleshooting workload. 13 model+backend combinations, 20 triage scenarios, LLM-as-judge scoring. Stay on llama.cpp confirmed; granite-4.1-8b-tuned emerged as the best speed-at-quality default.

Apple Silicon ASR Shootout: Picking a Transcription Engine

Mon, 01 Jan 0001 00:00:00 +0000

Four-engine reproducible benchmark of whisperx, parakeet-mlx, mlx-whisper, and whisper.cpp on Apple Silicon. 8 cohorts across 16 LibriSpeech clips, scored on WER and real-time factor. parakeet-mlx tdt-1.1b wins on the speed-at-quality frontier; mlx-whisper large-v3 disqualified itself on a hallucination loop.

Go বেসিক

Mon, 01 Jan 0001 00:00:00 +0000

Introduction

Mon, 01 Jan 0001 00:00:00 +0000

Hello World

A sample go program is show here.

package main

import "fmt"

func main() {
  message := greetMe("world")
  fmt.Println(message)
}

func greetMe(name string) string {
  return "Hello, " + name + "!"
}

Run the program as below:

$ go run hello.go

Variables

Normal Declaration:

var msg string
msg = "Hello"

Shortcut:

msg := "Hello"

Constants

const Phi = 1.618

Basic Types

Mon, 01 Jan 0001 00:00:00 +0000

Strings

str := "Hello"

Multiline string

str := `Multiline
string`

Numbers

Typical types

num := 3          // int
num := 3.         // float64
num := 3 + 4i     // complex128
num := byte('a')  // byte (alias for uint8)

Other Types

var u uint = 7        // uint (unsigned)
var p float32 = 22.7  // 32-bit float

Arrays

// var numbers [5]int
numbers := [...]int{0, 0, 0, 0, 0}

Pointers

func main () {
  b := *getPointer()
  fmt.Println("Value is", b)

func getPointer () (myPointer *int) {
  a := 234
  return &a

a := new(int)
*a = 234

Pointers point to a memory location of a variable. Go is fully garbage-collected.

অ্যাডভান্সড

Mon, 01 Jan 0001 00:00:00 +0000

Flow Control

Mon, 01 Jan 0001 00:00:00 +0000

Condition

if day == "sunday" || day == "saturday" {
  rest()
} else if day == "monday" && isTired() {
  groan()
} else {
  work()
}

if _, err := doThing(); err != nil {
  fmt.Println("Uh oh")

Switch

switch day {
  case "sunday":
    // cases don't "fall through" by default!
    fallthrough

  case "saturday":
    rest()

  default:
    work()
}

Loop

for count := 0; count <= 10; count++ {
  fmt.Println("My counter is at", count)
}

entry := []string{"Jack","John","Jones"}
for i, val := range entry {
  fmt.Printf("At position %d, the character %s is present\n", i, val)

n := 0
x := 42
for n != x {
  n := guess()
}

File Manipulation

Mon, 01 Jan 0001 00:00:00 +0000

Condition

if day == "sunday" || day == "saturday" {
  rest()
} else if day == "monday" && isTired() {
  groan()
} else {
  work()
}

if _, err := doThing(); err != nil {
  fmt.Println("Uh oh")

Bash Variables

Mon, 01 Jan 0001 00:00:00 +0000

Variable

NAME="John"
echo $NAME
echo "$NAME"
echo "${NAME}

Condition

if [[ -z "$string" ]]; then
  echo "String is empty"
elif [[ -n "$string" ]]; then
  echo "String is not empty"
fi

Autonomous Infrastructure Anomaly Detector (AIOps)

Mon, 01 Jan 0001 00:00:00 +0000

What: A predictive operations model that ingests live or historical server/cluster metrics to classify system health, proactively warning of node failures or memory leaks before orchestration panics.

Why: Bridges AI/ML coursework with existing infra skills (Talos, Debian, K8s). The dataset and problem are straightforward enough for a WGU project but real enough to be useful.

Approach: Python + Scikit-Learn classification pipeline. Historical metrics → binary classifier (normal vs. anomalous). Hyperparameter tuning, ensemble methods, regularization.

Go এর নোট সমূহ

Mon, 01 Jan 0001 00:00:00 +0000

Go Notes

Local LLM Inference Backend Eval: MLX vs llama.cpp on M1 Pro

Mon, 01 Jan 0001 00:00:00 +0000

Side-by-side evaluation of llama.cpp and MLX (via oMLX) for self-hosted inference on a single M1 Pro. Custom blind-scoring harness, 6 matched model pairs, 780 measurement rows. Zero migration recommended, and a silent audio bug surfaced in the MLX path.

Local LLM Performance & Throughput Predictor

Mon, 01 Jan 0001 00:00:00 +0000

What: A regression model trained on self-benchmarked hardware data that predicts tokens/second or VRAM consumption from inference parameters.

Why: Uniquely positioned. Already run local LLMs (Ollama, MLX) and have the hardware to generate real training data. The dataset is self-created, which strengthens the D683 proposal.

Approach: Python + Scikit-Learn regression. Scripted benchmarking loop across quantization levels, model sizes, context windows, and prompt lengths. Output: continuous TPS or TTFT prediction.

Data:

Smart Greenhouse & Hydroponic Climate Optimizer

Mon, 01 Jan 0001 00:00:00 +0000

What: An agricultural optimization model that parses environmental and nutritional vectors to predict crop growth health or determine optimal feeding schedules.

Why: Intersects with the aquaponics interest already on the radar (housekeeping tracker: “strongest energy thread is electronics + plants/aquaponics”). Turns a genuine hobby into coursework.

Approach: Python + Scikit-Learn. Tabular CSV data with environmental sensor readings. Either regression (predict growth rate) or classification (stress status).

Data:

Format: Tabular .csv with cultivation metrics
Features: Ambient temp, relative humidity, photoperiod hours, water EC/nutrient level, pH
Target: Continuous (biomass/growth rate mm) or categorical (stress T/F)
Cleanliness: Avoid raw images, tabular only. UCI Crop Recommendation dataset works out of the box.

Fit: D683 (custom dataset). UCI dataset is pre-cleaned, well-documented, and instantly approvable.

Curtis Nix

What I Run, and Why

Debian LUKS Locked Me Out of My Cluster for 14 Hours

A Power Outage Cut My Access To My Homelab During Vacation

Two-Board ESP32 Lab: ESP-NOW Foundations

Custom OTA over ESP-NOW

LLM Backend Eval Round 2: K8s Triage Scenarios

Apple Silicon ASR Shootout: Picking a Transcription Engine

Go বেসিক

Introduction

Hello World

Variables

Constants

Basic Types

Strings

Numbers

Arrays

Pointers

অ্যাডভান্সড

Flow Control

Condition

Switch

Loop

File Manipulation

Condition

Bash Variables

Variable

Condition

Autonomous Infrastructure Anomaly Detector (AIOps)

Go এর নোট সমূহ

Go Notes

Local LLM Inference Backend Eval: MLX vs llama.cpp on M1 Pro

Local LLM Performance & Throughput Predictor

Smart Greenhouse & Hydroponic Climate Optimizer

নোট সমূহ

ব্যাশের নোট সমূহ

Bash Notes