Hey there! I'm

Calvin DeBellis

Computer Science & Data Science — Let's create something awesome together.

Calvin DeBellis

About Me

Hey! I'm a UW–Madison graduate with degrees in Computer Science and Data Science. I've always been fascinated by computers and their near-limitless potential — and with the rise of AI, I'm all-in on data systems and what they can unlock.

From fine-tuning LLMs in Prague to building data pipelines and dashboards stateside, I love diving into tough problems with great people. I'm proficient in Python, SQL, and C, and I've worked with tools like Azure, Snowflake, Tableau, Docker, and more — always happy to connect and chat!

Education

Computer Science & Data Science

University of Wisconsin–Madison

Location

Madison, Wisconsin

Interests

AI/LLMs, Data Pipelines, Cloud & DevOps

Projects

TrickSki ML Classifier

Built a convolutional neural network to classify waterski trick orientation from custom video footage. Collaborated with a competition data provider for raw footage, designed the full dataset pipeline including image preprocessing and orientation-based spatial modeling, and optimized classification accuracy toward an automated judging system that can score a skier's run directly from video.

Python CNN Computer Vision

TrickSki Calculator

Details coming soon.

TBD

MiniSpark Engine

Implemented a multithreaded Spark-style data processing framework in C supporting map, filter, join, and partitionBy across partitioned datasets. Built a custom thread pool with POSIX threads and mutex/condvar synchronization, designed a recursive RDD materialization model with lazy computation and dependency tracking, and implemented partition caching to eliminate redundant computation under deep dependency chains.

C Multithreading Operating Systems

Spark Cluster Project

Built and deployed a distributed data processing cluster using Docker, coordinating multiple nodes for Spark computation and storage. Optimized query efficiency through data partitioning, caching, and execution plan analysis. Developed a machine learning model to predict loan approval outcomes and produced county-level visual analyses of loan trends and approval patterns.

Python Spark Docker ML

Experience

AI Intern

May 2025 — Aug 2025

Ataccama · Prague, Czechia · On-site

  • Researched fine-tuning strategies for LLMs to improve agent performance by reducing system prompt size.
  • Expanded and enhanced the agent evaluation framework to support iteration across multiple fine-tuned agents.
  • Identified potential to reduce system prompt token usage by up to 50% (11k → <6k).
  • Cut evaluation variance in half and increased unique evaluation questions by 140%.
  • Developed data gathering and scripting workflows for fine-tuning LLMs in Python.
  • Gained cloud experience with Azure and Amazon S3, including LLM fine-tuning in Azure AI Foundry.
  • Built data visualizations with Pandas, Plotly, and Matplotlib.
  • Worked with GitLab CI/CD pipelines, Docker, and Kubernetes.
  • Debugged GraphQL APIs by analyzing error logs.

Marketing Operations Intern

Jun 2024 — Sep 2024

Instructure · Salt Lake City · Remote

Over the course of this internship, I was tasked with creating a data pipeline for a Media Mix Model. This involved writing various SQL queries in Snowflake and navigating finance data from a different database. I used APIs in Google Sheets using Python to clean and aggregate data. I additionally learned about the statistics that go into a Media Mix Model from the data scientist developing it.

Revenue/Data Operations Intern

Jun 2023 — Aug 2023

Instructure · Salt Lake City · Hybrid

  • Worked with SQL, Salesforce, Tableau, and Excel to create Dashboards for the Revenue Operations team. I used data to accurately tell a story of what our team needs to keep doing and what they need to improve upon.
  • Worked in Salesforce cleaning, deduping, adding, and updating accounts.
  • Worked in multiple web scraping projects where I would gather intel off of school websites. This was to understand when server traffic would pick up based on the school's start dates.

Project Intern

Mar 2021 — Jun 2021

Inspirit AI · United States · Remote

Worked with a team developing an app to identify cancerous skin lesions from a photo. I specifically was in charge of data manipulation to enlarge training datasets for the model. I would often stretch a training set of a couple of photos by multiple factors through photo editing. I would make specific editing choices based off of accuracy vs. performance tradeoffs.

What People Say

Petr Dluhoš

Senior ML Engineer, Filevine · Supervising Engineer at Ataccama

“Calvin consistently delivered high-quality work with minimal supervision. He actively integrated into the team, sought out opportunities to learn, provided constructive feedback, and wasn't afraid to suggest improvements. The cooperation exceeded my expectations.”
“I supervised Calvin DeBellis during his internship at Ataccama in the summer of 2025. Throughout his time with us, Calvin demonstrated strong technical and analytical skills. He was able to take on a complex topic, break it down into manageable parts, and systematically work through each step. He approached his work with autonomy, clear communication of progress, and was not afraid to ask thoughtful and relevant questions when needed. Calvin successfully handled several technical challenges, including configuring our environments, connecting to cloud services, writing python scripts to extract data necessary for the project, and performing fine-tuning and evaluation of large language models in an agenting framework. He consistently delivered high-quality work with minimal supervision. Beyond his technical abilities and problem-solving mindset, I would like to emphasize Calvin's soft skills. He actively integrated into the team. He regularly attended team meetings, sought out opportunities to learn, provided constructive feedback, and wasn't afraid to suggest improvements — all of which were appreciated by the team. Overall, I enjoyed the cooperation with Calvin and it exceeded my expectations. Thus I can recommend him for any future academic or professional opportunities.”

Adam Drda

Engineering Manager · Supervisor at Ataccama

“From day one, Calvin showed strong initiative and a genuine drive to learn. He ran experiments, turned results into actionable insights, and contributed meaningfully to team discussions. I would gladly recommend him for roles in AI, data science, or engineering.”
“I had the pleasure of managing Calvin DeBellis during his internship on our AI team at Ataccama in the summer of 2025. From day one, Calvin showed strong initiative and a genuine drive to learn. His main focus was evaluating the impact of fine-tuning large language models and improving our internal evaluation framework, which plays an important role in how we measure model performance. He approached his work with curiosity, structure, and solid technical thinking. He ran experiments, turned results into actionable insights, and contributed meaningfully to team discussions. His ability to adapt quickly, communicate clearly, and collaborate with others made him a strong part of the team. I really enjoyed working with Calvin and would gladly recommend him for roles in AI, data science, or engineering.”

Jacquelyn Molino Ponton

Marketing Operations · Marketo Certified Expert, Instructure

“Calvin took full ownership of this project, demonstrating exceptional initiative and technical skill. He independently led meetings, showcasing impressive leadership abilities. Calvin's analytical mindset, problem-solving skills, and dedication to excellence made him an invaluable asset to our team.”
“I had the pleasure of managing Calvin during his summer internship at Instructure in 2024. During this time, he played a pivotal role in developing our new Marketing Media Mix model, which uses data science to determine optimal budget allocations across various marketing channels. Calvin took full ownership of this project, demonstrating exceptional initiative and technical skill. He worked directly with stakeholders at Instructure and a third-party company to define the scope of work and project requirements. He independently led meetings, showcasing impressive leadership abilities. Calvin was solely responsible for collecting and organizing data, skillfully setting up queries to consolidate all relevant information into a comprehensive table within our data warehouse. His attention to detail and methodical approach ensured that the data was not only accurate but also readily accessible for analysis. One of Calvin's standout contributions was his ability to troubleshoot issues as they arose. He identified a cumbersome manual process that was hindering our efficiency and successfully devised a solution using Python to automate it. This significantly improved the timeliness and accuracy of our data collection, enhancing the overall effectiveness of our Marketing Media Mix model. Calvin's analytical mindset, problem-solving skills, and dedication to excellence made him an invaluable asset to our team. I have no doubt that he will bring the same level of commitment and expertise to any future role and will be a tremendous asset to any organization.”

Quoc Tran

VP RevOps, Replit · Mentor at Instructure

“Calvin takes initiative to ask questions and broaden his views beyond just his work to understand business goals as a whole. His best qualities are his curiosity and ownership which lead him to a growth mindset. He continues to seek new opportunities for continuous improvement and deliver great results.”
“Calvin takes initiative to ask questions and broaden his views beyond just his work to understand business goals as a whole. I have observed that he has a strong work ethic and always takes advantage of additional learning opportunities. For example, Calvin proactively asked me to mentor him to explore different career paths based on my experience. To top it off, he works well with others inside his team, cross functionally, and outside of our company. Calvin is a well rounded employee with technical and social skills alike. I think his best qualities are his curiosity and ownership which lead him to a growth mindset. With that, he continues to seek new opportunities for continuous improvement and deliver great results.”

Corey McGarr

Director, Marketing Operations, Commerce

“I would highly recommend him for a data-focused role. Calvin worked with multiple teams and became proficient in data manipulation and analysis. He received high praises from all team leads who worked with him. Calvin has a very positive attitude towards his job.”
“I had the pleasure of managing Calvin over a summer internship at Instructure, and would highly recommend him for a data-focused role. During his time at Instructure, Calvin worked with multiple teams and became proficient in data manipulation and analysis. He used tools like Salesforce, Tableau, and Snowflake, and received high praises from all team leads who worked with him. Calvin has a very positive attitude towards his job, and was willing to dive into both fun and tedious projects alike.”

Get in Touch

I love meeting new people and hearing about exciting ideas. If you have a project, an opportunity, or just want to chat, don’t be a stranger!