August Vollbrecht

Data Scientist | Machine Learning | Cloud | Python

Building production‑ready pipelines and turning data into actionable insights. Specializing in NLP, predictive modeling, and scalable data architectures.

• Languages: Python, SQL, Java
• Data Science & ML: Pandas, NumPy, scikit‑learn, XGBoost, PyCaret, BERTopic, spaCy, Transformers, NLTK, SHAP
• Deep Learning: PyTorch, TensorFlow, Keras, Autoencoders
• Big Data: PySpark, Hadoop, SparkML, SparkDL
• Orchestration & MLOps: Apache Airflow, MLflow, Docker, GitHub Actions
• Cloud: AWS (S3, EC2, Lambda), Azure Data Factory, Azure Databricks
• Databases: PostgreSQL, Redis, SQLite
• APIs & Visualization: FastAPI, Uvicorn, Streamlit, Plotly, Matplotlib, Seaborn
• Testing & CI/CD: pytest, flake8, GitHub Actions
• Other: BeautifulSoup, Requests, statsmodels, scipy, horovod, Git, Jupyter Notebook, joblib, PyYAML

Production pipeline that ingests daily news and Reddit posts, detects emerging trends using BERTopic, and generates partnership recommendations. Orchestrated with Airflow, served via FastAPI, visualized with Streamlit.
Tech: Python, BERTopic, spaCy, Airflow, MLflow, Docker, GitHub Actions
Repository

End‑to‑end machine learning system for transaction fraud detection, featuring both Random Forest and deep learning autoencoder models. Includes automated retraining, drift monitoring, SHAP explainability, and a Streamlit dashboard. Deployed with CI/CD (GitHub Actions) and MLflow tracking. Tech: Python, scikit‑learn, PyTorch, SHAP, MLflow, Docker, GitHub Actions, Streamlit, pandas, numpy, seaborn, matplotlib Repository

Developed a predictive model (Random Forest / XGBoost) to identify the top 30% of customers most likely to open marketing emails. The model increased reach by 25% while maintaining send volume, validated through A/B testing and an automated retraining pipeline. Included hyperparameter tuning (GridSearchCV), MLflow tracking, and a six‑month simulation.
Tech: scikit‑learn, XGBoost, MLflow, pandas, matplotlib, seaborn, SHAP, FastAPI, Streamlit, GitHub Actions Repository

Built a multi‑agent system (Google ADK) that ingests Instagram data via CSV or Meta API, normalizes it, and applies LLM‑powered analytics (engagement rates, posting patterns, forecasting). Outputs BI‑ready datasets for PowerBI and Tableau with a Streamlit interface.
Tech: Python, Google ADK (Gemini), pandas, Streamlit, Meta Graph API
Repository

• General Assembly - Data Science Bootcamp
  Verification
• Microsoft - Build and Operate Machine Learning Solutions with Azure
  Verification
• Microsoft - Create Machine Learning Models in Microsoft Azure
  Verification
• Google - AI Agent Intensive Course
  Verification
• Microsoft - Azure Machine Learning for Data Scientists
  Verification
• Microsoft - Perform Data Science with Azure Databricks
  Verification
• IBM - Big Data with Spark and Hadoop Essentials
  Verification
• University of California, Davis - SQL for Data Science
  Verification

• LinkedIn: linkedin.com/in/august-vollbrecht
• Email: augustvollbrecht@proton.me
• GitHub: github.com/AugustAllYear