This is a submission for the New Year, New You Portfolio Challenge Presented by Google AI
Home - Learning Machine Learning in Public
This site documents my machine learning learning process through small, reproducible projects. Each step is deployed as a live demo and paired with clear explanations, mistakes, and reflections.
About Me
My name is Amankos Danial, currently learning data analysis and machine learning. I’m still at the beginner stage, so I’m intentionally keeping my projects small, honest, and well-explained. My goal is to become the kind of analyst who can turn messy data into clear insights and reliable predictions using every available instrument and observing patterns.
This portfolio is structured as a learning journal: what I tried, what worked, what didn’t, and what I learned. My goal is to learn as much as I possibly can from this, as well as to show clear thinking, steady progress, and respect for fundamentals.
I use Google AI tools, including Gemini, as a learning companion — to clarify concepts, validate my understanding, and reflect on results. All decisions, experiments, and conclusions are my own.
Credits
This portfolio documents my learning process, so transparency matters to me. Below are the tools, datasets, and resources I used while building and learning.
Libraries & Frameworks: Python, Flask, pandas, numpy, scikit-learn, matplotlib, seaborn.
Datasets: Breast Cancer Wisconsin dataset — provided directly by scikit-learn and used for beginner-friendly exploration and modeling.
Google AI Tools: Gemini (Google AI) — used as a learning companion to clarify machine learning concepts in simple language, help me interpret results and assist with drafting written reflections.
Infrastructure: Google Cloud Run — used to deploy and host this portfolio as a containerized application.
Portfolio
This portfolio documents my machine learning learning process through small, simple projects.
Each project focuses on understanding the fundamentals first — working with real datasets, building simple baseline models, and illustrating results clearly.
The portfolio itself is deployed on Google Cloud Run and embedded directly below.
How I Built It
I built this portfolio as a beginner-friendly ML learning log: each mini-step becomes a small, reproducible project that can be easily understood and explained.
Tech stack:
- Python + Flask for the web app
- pandas / numpy for data handling
- scikit-learn for baseline modeling (Logistic Regression and Decision Tree)
- matplotlib / seaborn for visualizations
Project structure:
- Project 1: dataset exploration (distributions + correlation heatmap)
- Project 2: baseline models with evaluation (accuracy + confusion matrices)
- Project 3: interpretation and reflection (feature importance for the Decision Tree)
Google AI tools:
I used Gemini as a learning companion: to clarify ML concepts in plain language, sanity-check my understanding of metrics (like confusion matrices), and help me draft reflections after each step. I still ran the experiments myself and verified results before writing anything.
Deployment
I containerized the app and deployed it to Google Cloud Run. The live site is embedded in this post, and the service was deployed with the required label:
--labels dev-tutorial=devnewyear2026.
What I'm Most Proud Of
- I shipped something real as a beginner. The portfolio isn’t a static template — it’s a working app with real outputs and real learning steps.
- I focused on fundamentals instead of difficult to understand functions and datasets. I didn’t jump into complicated models. I started with exploration, then baselines, then interpretation.
- I made the results explainable. Adding confusion matrices and feature importance helped me understand what the model is doing, not just how accurate it is.
- I deployed it properly. Getting everything live on Cloud Run (and embedded in Dev.to) taught me a lot about real-world deployment and reliability.
Top comments (0)