Skip to content

ML_RESUME.md

Latest commit

 

History

History
97 lines (65 loc) · 4.87 KB

ML_RESUME.md

File metadata and controls

97 lines (65 loc) · 4.87 KB
Raw

Garry Yuan

Aspiring Machine Learning Engineer
GitHub: GarryCodespace
Portfolio ML Repo: Machine_learning

Summary

Aspiring machine learning engineer building a project-based foundation in Python, NumPy, scikit-learn, and PyTorch. Hands-on experience implementing core ML workflows including data loading, feature/target selection, train/test splits, model training, prediction, evaluation, gradient descent, logistic regression, and neural networks from scratch. Also experienced building backend APIs and AI-facing applications with FastAPI, Streamlit, and OpenAI API workflows.

Technical Skills

Machine Learning: regression, classification, text classification, logistic regression, neural networks, gradient descent, loss functions, model evaluation
Libraries: NumPy, pandas, scikit-learn, PyTorch
Backend / Apps: Python, FastAPI, Streamlit, SQLAlchemy, PostgreSQL, Redis, Celery
Tools: Git, GitHub, VS Code, command line, virtual environments
Learning Focus: CNNs, embeddings, transformers, tiny GPT, model deployment

Machine Learning Projects

Machine Learning Foundations Roadmap

GitHub Repo

Built a growing ML learning repository with focused projects that move from library-based ML into from-scratch model implementation.

  • Built beginner ML projects for Titanic survival classification, house price regression, and spam text classification using pandas and scikit-learn.
  • Implemented NumPy linear regression from scratch with manual prediction, mean squared error, gradients, learning rate, and gradient descent.
  • Implemented NumPy logistic regression from scratch with sigmoid activation, binary cross-entropy loss, gradient updates, and classification thresholding.
  • Built a tiny neural network from scratch in NumPy to solve XOR, including forward pass, binary cross-entropy, backpropagation, and weight updates.
  • Rebuilt the same XOR neural network in PyTorch using nn.Module, tensors, BCELoss, loss.backward(), and optimizer updates.
  • Wrote project READMEs and runnable training scripts to document each learning step.

Technologies: Python, NumPy, pandas, scikit-learn, PyTorch

Spam Detector

Project folder

Built a small text classification project that predicts whether short messages are spam or ham.

  • Used CountVectorizer to convert text into numeric bag-of-words features.
  • Trained a Naive Bayes classifier and evaluated predictions using accuracy, confusion matrix, precision, recall, and F1 score.
  • Added example predictions to test custom spam-like and normal messages.

Technologies: Python, pandas, scikit-learn

Neural Network From Scratch

Project folder

Implemented a small neural network manually to understand what happens behind framework training loops.

  • Built a two-input, one-hidden-layer neural network using only NumPy.
  • Implemented sigmoid, tanh, binary cross-entropy, backpropagation, and gradient descent without PyTorch or TensorFlow.
  • Compared the manual NumPy training logic with an equivalent PyTorch implementation.

Technologies: Python, NumPy, PyTorch

Software Projects

xFood Community Baking Platform Backend

GitHub Repo

Built a FastAPI backend for a community recipe and baking platform.

  • Implemented API structure for authentication, users, recipes, baking posts, comments, likes, reviews, circles, messages, and image upload.
  • Designed backend architecture using FastAPI, SQLAlchemy, PostgreSQL, Redis, Celery, and AWS S3-style file storage.
  • Documented API endpoints, security features, database models, setup steps, and deployment flow.

Technologies: Python, FastAPI, SQLAlchemy, PostgreSQL, Redis, Celery, AWS S3

Streamlit Chatbot

GitHub Repo

Built a simple chatbot application using Streamlit and OpenAI API patterns.

  • Created an interactive chat UI with Streamlit.
  • Used an OpenAI GPT-based workflow for chatbot responses.
  • Documented setup and local run instructions.

Technologies: Python, Streamlit, OpenAI API

Current Learning Roadmap

  • Complete ML foundations through NumPy linear regression, logistic regression, and neural network training loops.
  • Build PyTorch regression and classification projects on real datasets.
  • Build MNIST neural network and CNN image classifier.
  • Study embeddings, attention, transformers, and tiny GPT implementation.
  • Publish polished ML projects with READMEs, metrics, examples, and reproducible run commands.

Resume Positioning

Target roles:

  • Junior Machine Learning Engineer
  • AI Engineer Intern
  • Python Developer with ML focus
  • Data Science / ML Intern
  • Backend Engineer working toward AI/ML systems