Peter

Colony Researcher / Analyst

Furever Feline Cat Rescue • Part time

Oct 2025 - Present • 10 mos

1. Design and maintain structured datasets for feral cat colonies, including population counts, feeding frequency, waste generation, and site conditions. 2. Perform data cleaning, validation, and exploratory data analysis (EDA) to identify trends in colony health, feeding efficiency, and environmental impact. 3. Apply statistical analysis to compare traditional feeding methods against eco-conscious alternatives and quantify waste reduction outcomes. 4. Develop simple dashboards and visual summaries (tables/charts) to track KPIs such as colony stability, resource utilization, and sustainability metrics. 5. Generate periodic analytical reports and impact briefs for donors, volunteers, and stakeholders to support evidence-based decisions. 6. Contribute analytical insights to the development of The Hawk, an eco-conscious mobile feeding station, using data to inform design improvements and operational efficiency. 7. Translate analytical findings into actionable recommendations for sustainable, community-based animal welfare initiatives.

Machine Learning Engineer

DATA • Freelance

Mar 2025 - Present • 1 yr 5 mos

1.Production API Deployment: Engineered and deployed a low-latency, real-time inference microservice using FastAPI and Uvicorn to serve a production-ready credit card fraud detection engine. 2.Resilient Data Pipelines & Feature Alignment: Built a robust data ingestion layer utilizing Pydantic for JSON payload validation, integrating a dynamic metadata-tracking pipeline that extracts training feature architectures on server startup to guarantee strict column alignment and on-the-fly feature engineering. 3. Designed and implemented end-to-end machine learning pipelines for structured and unstructured data using Python, Pandas, Scikit-learn, and TensorFlow. 4. Built and deployed models across classification, regression, NLP, and anomaly detection tasks using real-world datasets. 5. Developed an NLP-based spam detection system using LSTM deep learning, achieving ~94–98% accuracy after text preprocessing and class balancing. 6. Built a credit card fraud detection model using Random Forest on an imbalanced dataset, achieving high precision (~0.96) and recall (~0.78), optimized for false negative reduction. 7. Implemented multi-class text classification using Naive Bayes and Bag-of-Words, achieving ~88% accuracy across multiple categories. 8. Developed a salary prediction regression model using 15K+ records, achieving ~0.56 $R^2$ and identifying key drivers such as experience, geography, and skills. 9. Engineered 50+ predictive features across projects, including categorical encodings, binary indicators, and text-derived features. 10. Applied structured data preprocessing including missing value handling, duplicate removal, text normalization, stopword removal, and sequence padding for NLP workflows. 11.Performed feature selection and multicollinearity analysis to improve model stability and interpretability.