Coding Projects

Stock Analysis & Grading Model App

Stock analysis app using a custom-built grading model optimized to evaluate primarily on growth-based equity assets that would be held within a 6-12 month period. Home page previous day data to assign grade, while other pages uses grades that are batched on a weekly basis via Github Actions orchestration; stored in S3. The grading model point system can be found in my github repo. Metrics/histiorical prices pulled from yfinance lib.

PythonStreamlitGitHub ActionsPandasNumPyMatplotlibScikit-learnPlotlySeabornAWS S3Monte Carloyfinance

GitHubApp URL

Delivery Route Optimization App

Built for Brightside non-profit in Minneapolis MN for their PWYC delivery service to automate the route assignment and distribution process. The app uses a 2-step categorization process that uses machine learning to group routes into geographic cluster then orders and redistributes routes into more optimally balanced groupings based on travel/traffic times by car & process-specific hardcoding. The link below is an alternate version for viewing/testing.

PythonStreamlitPandasScikit-learnK-MeansGoogle APIAWS S3

GitHubApp URL

AWS Athena AI Agent

AI agent interface that allows users to query AWS Athena using natural language. The agent uses AWS Bedrock for secure, enterprise-level AI interactions and converts natural language questions into SQL queries (Prompt-based Semantic Parsing), executes them against Athena, and provides results with analytical summarizations.

PythonAWS AthenaAWS BedrockSemantic ParsingPrompt EngineeringPandasGoogle APIAWS S3

GitHubApp URL

Private Credit Solutions Mock ETL & Medallion Master Data Architecture

A mock data architecture for private credit lending, demonstrating a domain-oriented data engineering pipeline built with dbt and Apache Airflow. The architecture follows a medallion-style ingestion and transformation framework, progressing from raw source data through structured and curated domain layers to an analytical output layer. Industry-standard MDM principles (mastered into data domains) are applied at the curated layer to produce clean, governed master data entities. DuckDB is used as the local test target with Snowflake as the production environment, with dbt models handling dialect-switching between the two.

PythondbtAirflowDuckDBSnowflakeMaster Data ManagementData ModelingData OrchestrationETL Pipelines

GitHub