Kevin Della Piazza

As an AI/MLOps Engineer and Cloud Architect, I specialize in one thing:

building end-to-end, production-grade systems on the cloud.

My portfolio is the tangible proof of my core philosophy (Automation, Serverless-First, Cost-Optimization) in action:

𝗣𝗿𝗼𝗷𝗲𝗰𝘁 𝟭: 𝗥𝗲𝘃𝗶𝗲𝘄𝗟𝗲𝗻𝘀 𝗔𝗜 (𝗦𝗲𝗿𝘃𝗲𝗿𝗹𝗲𝘀𝘀 𝗠𝗟𝗢𝗽𝘀)

• Engineered a multi-layer AI pipeline that processes customer reviews, transforming raw feedback into actionable insights, at an estimated cost of ~$9 per million reviews.

• Built an end-to-end pipeline using Terraform (IaC) and a Step Functions "fan-out" architecture.

• Solved AWS's 10GB size & dependency limits by containerizing 4 Hugging Face models with Docker Lambdas.

𝗣𝗿𝗼𝗷𝗲𝗰𝘁 𝟮: 𝗩𝗖𝗠 - 𝗩𝗔𝗧 𝗖𝗼𝗺𝗽𝗹𝗶𝗮𝗻𝗰𝗲 𝗠𝗼𝗻𝗶𝘁𝗼𝗿 (𝗙𝗶𝗻𝗧𝗲𝗰𝗵 𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻)

• Designed a serverless pipeline that processes invoices at scale, creating a fully auditable and compliant data trail for FinTech, at an estimated cost of ~$18 per 10,000 invoices.

• Automated the workflow with a 100% serverless, AWS SAM (IaC) pipeline

• Leveraged AWS Textract (intelligent OCR) to extract data from scanned PDFs and AWS Glue/Athena to make it queryable.

𝗣𝗿𝗼𝗷𝗲𝗰𝘁 𝟯: 𝗗𝗮𝘁𝗮 𝗙𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝘀 𝗙𝗼𝗿 𝗔𝗜 (𝗦𝘁𝗿𝗮𝘁𝗲𝗴𝗶𝗰 𝗘𝗟𝗧)

• This project captures my strategic playbook for Data Governance, turning chaotic data silos into a governed "Single Source of Truth."

• Implemented a repeatable ELT process using the Medallion Architecture (Bronze/Silver/Gold) to ensure data quality.

• Designed analytics-optimized Star Schemas to accelerate BI dashboards and prepare data for AI models.

I co-founded a 3-person FinTech venture, Tortuga MKM, where we focused on turning complex raw data into a real competitive advantage.

As a founder, my role combined the responsibilities of a core business owner and our Lead Data Engineer.

I was responsible for architecting and building our entire data operation from the ground up.

This was my journey:

• I began by managing our analysis via manual processes in Excel, but quickly understood its limitations when faced with thousands of raw, time-series data points from financial markets. Recognizing this inefficiency as a 𝗰𝗿𝗶𝘁𝗶𝗰𝗮𝗹 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗿𝗶𝘀𝗸, I took the initiative to re-engineer our entire workflow to a more robust and automated stack using Python and SQL.

• In this period, I realized that trustworthy data is the foundation of any successful business. I designed the system not just to move data, but to ensure its integrity and reliability at every stage while also enabling clear data visualization to drive action.

Results:

The impact of this automation 𝘀𝗹𝗮𝘀𝗵𝗲𝗱 𝗼𝘂𝗿 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀 𝗰𝘆𝗰𝗹𝗲 𝘁𝗶𝗺𝗲 𝗯𝘆 𝗼𝘃𝗲𝗿 𝟵𝟬%, taking what was a multi-day manual process into an automated workflow that ran in hours. This efficiency 𝗲𝗹𝗶𝗺𝗶𝗻𝗮𝘁𝗲𝗱 𝗺𝗮𝗻𝘂𝗮𝗹 𝗲𝗿𝗿𝗼𝗿𝘀 and 𝗯𝗲𝗰𝗮𝗺𝗲 𝗼𝘂𝗿 𝗽𝗿𝗶𝗺𝗮𝗿𝘆 𝗰𝗼𝗺𝗽𝗲𝘁𝗶𝘁𝗶𝘃𝗲 𝗮𝗱𝘃𝗮𝗻𝘁𝗮𝗴𝗲, directly enabling our team to execute on complex, data-driven strategies that were impossible with our old manual process.

This foundational journey of building a business-critical system out of pure necessity ignited my passion for data engineering.

It was the perfect launching point that set my next strategic goal: to master the cloud architecture required to build the end-to-end, production-grade systems I specialize in today.