Mood Board v1: Initial

Electric Modern aesthetic • Bright Red, Cyan, Lime • Inter Typography

Color Palette

Foundation

Canvas#FAFAFA
Surface#FFFFFF
Border Light#E5E7EB
Border#D1D5DB

Accent Colors

Bright Red (Primary)

Light#FEE2E2
Default#EF4444
Dark#DC2626

Cyan (Secondary)

Light#CFFAFE
Default#06B6D4
Dark#0891B2

Lime (Tertiary)

Light#ECFCCB
Default#84CC16
Dark#65A30D

Text Colors

Black#000000
Black Accent#1A1A1A
Text Primary#1A1A1A
Text Secondary#6B7280

Typography

Heading 1 - The Art of Building ML Systems

Heading 2 - Project Showcase

Heading 3 - Technical Deep Dive

Heading 4 - Implementation Details

Heading 5 - Code Examples
Heading 6 - Notes & References

Body XL: After 15+ years of building production ML systems, I've learned that the hardest part isn't the algorithms—it's everything else.

Body Large: From firmware-level signal processing to cloud-scale ML pipelines, I work at the intersection of research and product.

Body: The goal is always the same: ship systems that actually work in the real world, not just in notebooks. This means thinking about deployment, monitoring, data quality, and user experience from day one.

Body Small: Technical depth with big-picture thinking • Pragmatic innovation • Genuine curiosity

Inline code example: const result = await processData()

function trainModel(data: Dataset): Model {
  // Kalman filtering for state estimation
  const filtered = applyKalmanFilter(data);

  // Feature engineering
  const features = extractFeatures(filtered);

  return model.train(features);
}

Buttons

Variants

Sizes

States

Form Components

Contact Form Example
Demonstrating form inputs with our design system

Card Components

FLUID Water Meter
IoT hardware + ML for leak detection

Designed ultrasonic flow meter with real-time signal processing and ML-based anomaly detection for residential water monitoring.

Signal Processing
Machine Learning
Cloud ML Pipelines
Production-scale data processing

Built distributed ML training pipelines handling millions of data points, with automatic versioning and deployment.

Cloud Architecture
MLOps
Research to Product
Bridging the gap that matters

Translating cutting-edge research into production systems that ship. The hard part is making algorithms work at scale.

Innovation
Product

Andrew Aarestad

AI/ML Engineer • 15+ Years Shipping Production Systems

From firmware-level signal processing to cloud-scale ML pipelines, I build systems that bridge research and product. Technical depth with big-picture thinking.

Building Production ML Systems: Lessons from 15 Years

12 min read
Machine Learning
Engineering

After building ML systems for over 15 years, I've learned that the hardest part isn't the algorithms—it's everything else. Here's what they don't teach you in research papers.

The Data Pipeline is Everything

Your model is only as good as your data pipeline. I've seen brilliant algorithms fail in production because the data infrastructure couldn't keep up. Before you write a single line of ML code, ask yourself:

  • How will data flow from source to model?
  • What happens when upstream systems change?
  • How do you detect data quality issues?
  • Can you version your datasets?

Key Insight

Spend 80% of your time on data infrastructure, 20% on models. The best algorithm can't fix bad data.

Monitoring Matters More Than Metrics

Accuracy on your test set is nice. Knowing your model is degrading in production is critical. Real-world data drifts, user behavior changes, and upstream systems break. Your monitoring needs to catch:

// Example monitoring check
if (predictionConfidence < THRESHOLD) {
  alert("Model confidence dropping");
  logFeatureDistribution();
  checkDataDrift();
}

Set up alerts for distribution shifts, prediction confidence, latency spikes, and error rates. Better to know about problems before your users do.

Deployment is Not the Finish Line

Shipping v1 is just the beginning. Plan for:

  • Model retraining cadence
  • A/B testing framework
  • Rollback procedures
  • Feature store management
  • Feedback loops from users

This is part of a series on practical ML engineering. Next up: debugging production ML systems.

Data Tables

Project Timeline
Notable ML projects over the years
A timeline of production ML systems
YearProjectDomainImpact
2024Generative AI Platform
NLP
1M+ generations
2022FLUID Water Meter
IoT + ML
Patent filed
2020Anomaly Detection System
Time Series
99.9% uptime
2018Computer Vision Pipeline
CV
100K images/day
2015Recommendation Engine
RecSys
+40% engagement

Embedded Media

Video Example
Demonstrating embedded video with responsive aspect ratio
🎬

Video Embed Placeholder

16:9 responsive aspect ratio

aspect-video

Replace this placeholder with an iframe or video element. Example: <iframe src="..." />

Design System

Electric Modern aesthetic with bold accent colors and Inter typography. Built with Next.js and Tailwind CSS.

Colors

  • Bright Red #EF4444
  • Cyan #06B6D4
  • Lime #84CC16
  • Canvas #FAFAFA

Typography

  • Headings: Inter (500-800)
  • Body: Inter (400-500)
  • Code: JetBrains Mono

Mood Board v1: Initial • Created January 2025