The Final Algorithm: My Year of ML, MLOps, Agents and AI Alignment
I'm going to document my growth, insights, and lessons I learn, in the rapidly evolving world of AI.
Maybe I’m drinking too much of the AI-laced Kool-Aid?
Yeah. There’s no doubt that that is exactly what I’m doing.
Because when I get that positive feedback from ChatGPT, Claude, or Gemini, it triggers those dopamine receptors just right. With closing phrases like “Great idea!” or “You’re on track…” or “Keep working on this, because you are on to something…” make me feel like I’m on top of the world. Which leads to the unfortunate side of this: AI-generated responses are addictively encouraging.
Encouraging enough to help me pop this introvert bubble and start a year-long journey to improve my knowledge of AI—publicly. I'll show where I'm at as I journal my interests and the things I learn along the way. That will not only keep me honest with my goals—but I feel like it may help shame me into sticking to what I start. In a nutshell—it will help me from slacking off with this one. I’ve got to follow this passion to the end.
Hell, I thought about doing all of this on YouTube, but I'm not an on-camera kind of person. I love to write, so here goes...
This is The Final Algorithm. My next year-long journey into everything AI and how I will use it to improve my life, work, and—hopefully—help others along the way.
But I’m not stopping at just learning the technical skills like ML, MLOps or creating specialized AI agents. I’m also diving into something deeper: exploring the ethical and philosophical challenges of AI Alignment. What happens when the systems we build don’t inherently align with human goals—or worse, when they unknowingly act in ways that lead to harmful zero-sum outcomes? How do we ensure that the DNA of these systems accounts for interdependence, cooperation, and safeguards against runaway optimization? And yes — I wrote this. Not ChatGPT.
Basically, these questions are what fascinate me most about AI—not just the “how” of building it, but the “why” and “what if.” If AI agents are going to play a bigger role in our lives, then understanding how they reason, decide, and act is pretty critical. At least I think they are. Can we ensure they’re not just powerful but also responsible? And what can someone like me—an average person learning this step by step—bring to that conversation?
I personally don’t have all the answers yet, but that’s what this year-long journey (and maybe more) is for: to learn, to ask better questions, and maybe even propose some common-sense ideas that could possibly make a difference, somewhere, some place down the line.
I’ll add anecdotes along the way and test a few additional learning paths based on inspirations that I’m getting from watching Santiago (YouTube @underfitted).
First stop — 3 Blue 1 Brown — I’ve got some Maths to do.
BUT—before I do that, let me just share the year-long roadmap, so you can see what “kind of crazy” I’ve put on my shoulders…now here is what ChatGPT offered:
Months 1-2: Foundational Knowledge
Goal: Build foundational understanding of ML concepts and tools.
Research the Field:
Read about ML workflows and tools.
Recommended Resource: "Machine Learning Engineering" by Andriy Burkov (arriving in paperback - Jan 7th, 2025)
Watch beginner-friendly videos on ML applications (e.g., Google’s AI channel, and FreeCodeCamp).
Brush Up on Math:
Study linear algebra, calculus, probability, and statistics using:
Khan Academy for fundamentals.
3Blue1Brown YouTube series for intuitive understanding.
Learn Python for ML:
Reinforce Python skills (I’m doing a brush up on LinkedIn Learning now).
Learn libraries: NumPy, Pandas, Matplotlib, and Seaborn.
Practice by solving small problems on Kaggle or LeetCode.
Months 3-4: Data Skills and Exploratory Data Analysis (EDA)
Goal: Learn to work with data effectively.
Master Data Manipulation:
Get hands-on with Pandas for data cleaning and analysis.
Use Jupyter Notebooks to document your workflows.
Practice SQL:
Learn to query, join, and analyze data with SQL.
Recommended Course: Mode Analytics SQL Tutorial.
Focus on EDA:
Explore datasets to find patterns, visualize data, and detect anomalies.
Tools: Matplotlib, Seaborn, Plotly.
Mini Projects:
Example: Analyze a Kaggle dataset (e.g., COVID data or customer trends) and create a report with key insights.
Months 5-6: Machine Learning Basics
Goal: Learn core ML algorithms and apply them to real-world datasets.
Understand ML Theory:
Focus on:
Supervised Learning: Linear regression, logistic regression, decision trees.
Unsupervised Learning: Clustering, PCA.
Recommended Course: Andrew Ng’s ML Course on Coursera.
Learn Scikit-Learn:
Build ML models using Scikit-learn.
Practice cross-validation, feature engineering, and hyperparameter tuning.
Projects:
Example 1: Predict house prices using regression.
Example 2: Classify emails as spam or not spam.
Months 7-8: Deep Learning and Specialized ML Skills
Goal: Gain foundational skills in neural networks and explore advanced ML concepts.
Learn Neural Networks:
Study perceptrons, activation functions, backpropagation, and optimization.
Recommended Course: Deep Learning Specialization by Andrew Ng.
Use TensorFlow or PyTorch:
Build and train deep learning models (e.g., MNIST digit classification).
Explore Transfer Learning:
Use pre-trained models for tasks like image recognition or NLP.
Experiment with Real-World Applications:
Example Projects:
Sentiment analysis on customer reviews.
Image classification with transfer learning.
Months 9-10: MLOps and End-to-End Projects
Goal: Learn to deploy and monitor ML models in production.
Study MLOps Concepts:
Understand CI/CD pipelines, model versioning, and deployment workflows.
Tools: MLflow, Kubeflow, or TFX.
Learn Cloud Platforms:
Focus on AWS, GCP, or Azure for ML model hosting.
Practice deploying a model using Flask, FastAPI, or Streamlit.
End-to-End Project:
Example: Build a recommendation system or a fraud detection pipeline.
Steps: Data preprocessing → Model training → Deployment → Monitoring.
Months 11-12: Portfolio, Networking, and Continued Growth
Goal: Showcase your skills, expand your professional network, and continue learning to stay ahead of an emerging field.
Build Portfolio Projects:
Focus on 2-3 polished projects that demonstrate:
EDA, model training, and deployment skills.
Example: A dashboard showing predictions or insights.
Polish Your Online Presence:
Update your LinkedIn and GitHub profiles to highlight:
ML skills, certifications, and project achievements.
Contributions to open-source ML or related initiatives.
Network Strategically:
Join relevant LinkedIn groups, attend meetups, and engage in ML-focused communities.
Collaborate with others on open-source ML projects to broaden your expertise.
Continued Learning:
Stay current with advancements in ML by:
Following industry blogs, podcasts, and conferences.
Exploring emerging tools and techniques, such as explainable AI (XAI) or generative AI.
Professional Development:
Explore opportunities to apply ML concepts in your current role or field.
Leverage your growing expertise to provide thought leadership, share insights, or mentor others in ML basics.
Key Considerations
Time Commitment: Dedicate 10-15 hours per week (adjust as needed based on work-life balance).
Consistency: Break monthly goals into weekly objectives.
Flexibility: Adapt the roadmap as your interests and progress evolve.
Focus on Your Strengths: Your experience in software engineering and testing is a unique asset that you can leverage in roles like MLOps or applied ML engineering.