Andrew Castro

Andrew Castro

Data Engineer

Languages & Libraries

Python (+Pandas, Numpy, scikit-learn), SQL, PHP


Databases & IDEs

Microsoft SQL Server, MySQL, VS Code, Jupyter Notebook


Tools

Power BI, Excel, Microsoft 365


Skills

SWE, LLMs, exploratory analysis, data pipelines, databases, regression modeling, forecasting, visualization

ABOUT ME

I'm Andrew, a Data Engineer based in the United States, dedicated to finding clarity in complexity. As a first generation college graudate (B.S. in Software Engineering, 3.7 GPA, Honors), and second generation citizen in my family; I am driven to pursue a career with a positive global impact.

My passion is exploring analytics. I enjoy researching metrics for patterns and complex narratives to enable clear, actionable insights. I believe data is a powerful tool for shaping ethical and equitable policies, for everyone.

I am driven to grow my skills and apply them to high-impact fields, including biotechnical, machine-learning, and quantum engineering.

EMPLOYMENT

  • Website
  • Strategized a production-ready commercial platform end-to-end, translating business needs into a fully deployed system. 14
  • Engineered a PII-compliant appointment system using Google Cloud tools, mitigating data processing liability, while enabling the capture of all appointment-driven inbound leads. 2
  • Built a serverless architecture on Cloudflare Pages, cutting hosting and maintenance costs by $650/year while enabling modular backend expansion. 1

PROJECTS

  • Github Portfolio Website
  • Orchestrated asynchronous data ingestion from Open-Meteo, chaining Geocoding and Weather endpoints to reduce frontend latency by 40%. 14
  • Implemented Pydantic data models, enforcing strict schema validation to ensure type-safe JSON responses and eliminate runtime type errors. 2
  • Engineered a fuzzy-logic filtering algorithm to handle ambiguous user inputs (e.g., Warwick, RI vs Warwick, UK), improving location accuracy. 1
  • Deployed to a cloud environment (Render), configuring CORS policies and Gunicorn/Uvicorn workers for production accessibility. 1
  • Designed a custom scoring logic that normalizes temperature, wind, and rain data into a 0-100 "Score Index." 1

  • Github Portfolio Website
  • Engineered an end-to-end analytics pipeline to process and forecast FBI NIBRS crime statistics, across 9 years of multi-state datasets, with 91.69% forecasting accuracy in Total Offenses. 14
  • Scripted XLS to CSV conversion and normalized malformed headers across annual archives. 2
  • Merged datasets into a unified analytical model supporting multi-year trend analysis. 1
  • Developed forecasting models with MAE-based accuracy evaluation, revealing Autoregression improved error magnitude by 96.58% over Linear Regression. 1
  • Designed a 4-page Power BI report visualizing forecasts, residuals, and model error distributions. 1

  • Github Portfolio Website
  • Analyzed 100,000+ medical and lifestyle records to identify demographic, socioeconomic, and behavioral predictors of diabetes. 1
  • Engineered custom DAX measures to compare employment and income-level patterns, revealing near-identical positivity rates between employed (39.85%) and unemployed (39.78%) groups. 1
  • Identified that 70% of diabetes-positive patients lacked family history, and quantified average positive patient markers (PP glucose 160.04 mg/dL; triglycerides 123.21 mg/dL). 1
  • Highlighted BMI and LDL cholesterol as leading risk factors, supporting data-driven intervention strategies. 1

  • Github
  • Built a Power BI dashboard analyzing 34,000+ sales transactions, identifying seasonal profit peaks (+13.64% in Q3) and major loss-driving categories (e.g., groceries at –29.68%). 1
  • Found strong correlation between extended delivery times and product returns, revealing a root cause of margin leakage. 1
  • Provided actionable recommendations including logistics optimization, targeted promotions, and category-level margin restructuring. 1

  • Github
  • Developed a full-stack PHP collaboration platform supporting user authentication, role-based access control, and secure CRUD operations across users, groups, tasks, and messaging using PDO prepared statements and transaction-safe workflows. 1
  • Designed and optimized complex SQL joins, dynamic search filters, and conditional query builders, enabling fast retrieval of tasks, messages, and user/group metadata while maintaining strict moderator vs. member visibility rules. 1
  • Implemented a task management engine with update, completion, scheduling, and history-tracking features, including permission-scoped deletion logic and automatic status transitions. 1
  • Built a threaded messaging and memo system with conversation grouping, reply-state tracking, hide-without-delete functionality, and parent-child message relationships for clean inbox workflows. 1
  • Created modular backend components for advanced features such as automatic group color assignment, invite-code Bubble onboarding, sale/transaction confirmations, and user rating aggregation with fraud prevention safeguards. 1
  • Improved system reliability by structuring code into cohesive, scenario-based methods, reducing duplication and simplifying future maintenance across 30+ backend operations. 1

  • Github
  • Developed a full-stack eBay-style marketplace implementing user registration, login, authentication, and profile management using secure, database-driven workflows. 1
  • Built a product listing system that supports multi-image uploads, categories, item conditions, pricing, and location, enabling end-to-end item creation similar to eBay’s listing flow. 1
  • Implemented an advanced search engine with filters for keywords, description, category, condition, state, and seller, providing highly relevant results and improving user navigation across large inventories. 1
  • Created a buyer–seller messaging module enabling threaded conversations, unread status tracking, and inbox/outbox views to support negotiation and transaction coordination. 1
  • Designed moderator/admin dashboards for managing users, categories, conditions, and system-wide data, enabling complete CRUD control and safe administrative actions. 1
  • Integrated a sale confirmation workflow including order ID generation, time-stamped confirmation messages, and automated validation, improving transaction clarity and reliability. 1
  • Optimized backend logic with structured, scenario-specific methods and normalized database schemas, simplifying maintenance and reducing code duplication across the platform. 1