Daniel Rasmussen Email & Phone Number

Professional SummaryData scientist with a PhD in Earth Science. Over 7 years of experience building data-driven models to solve problems and gain insights using programming, statistics, and machine learning. Background in project design and scoping, reducing instrument data, and drawing insight from noisy, multidimensional data. Skilled at distilling insights (and identifying questions) from large volumes of information. Passion for finding solutions to societally relevant problems. Avid learner and strong team player.Skills⦁ Object-oriented programming and relational databases (Python | SQL)⦁ Data wrangling and analysis (PyData stack | GeoPandas)⦁ Predictive modeling, including regression, classification, and clustering (Scikit-learn | NumPy | SciPy)⦁ Ensemble learning (XGBoost | Gradient Boosting | Random Forests | Decision Trees)⦁ Statistics and survival analysis (Statsmodels | Scikit-survival | Lifelines)⦁ Web APIs (Flask | SQLAlchemy)⦁ Visualization and dashboarding (Matplotlib | Plotly | Seaborn | Altair | Streamlit)⦁ Problem solving and ability to distill insights from large volumes of dataIndustry Knowledge⦁ Structured and unstructured data (Geochemical | Geospatial | Text | Marketing)⦁ Data communication to stakeholders (non-technical and technical)⦁ Web scrapping (Web APIs | Requests | Beautifulsoup | Selenium)⦁ Git (GitHub | command line)⦁ Monte Carlo simulation and uncertainty analysis⦁ A/B testing⦁ Survival (time-to-event) analysis⦁ Technical consultingExample recent projects(1) Sustain-table: A web application, built using Flask and SQLAlchemy, to help decarbonize your diet while simplifying the process of creating grocery lists. You can build a repertoire of recipes using their urls. Information on the carbon emissions of each recipe is provided, and intelligent ingredient substitutions are offered to lower carbon emissions using a hybrid expert/ML(word2vec) model I built.(2) Probabilistic forecasting of volcanic eruption duration: Implemented machine learning (e.g., XGBoost, Random Forests) and traditional approaches to multivariate survival analysis, improving the discriminatory power of eruption models by ≤38%. Mined data from 17K eruptions (NLP).(3) Predicting contact behavior: Created models to analyze email contact list behavior and improve email deliverability, engagement, and donation revenue. Developed A/B testing methods with multi-armed bandit approach. Improved open rates by 15% to 30% with clustering (e.g., k-means) and classification (e.g., XGBoost) models. Helped campaign reach spending cap.