Nicholas
McCracken.

Software Engineer • AI/ML • Data Engineering

I'm a software engineer and M.S. CSE student at Ohio State with experience across backend systems, data engineering, cloud infrastructure, and AI/ML.

Choose a landmark to explore ✧ A FIELD ATLAS · FIVE LANDMARKS ✧
Graduating December 2026 • Columbus, OHSCROLL TO EXPLORE
01 PAST ADVENTURES

Experience.

May 2026 – Aug. 2026

AI & MACHINE LEARNING

Nationwide

AI/ML Engineering Intern

Columbus, OH

  • Built a distributed claims risk pipeline on AWS with PySpark, engineering features from claim severity, geography, and historical data to improve model accuracy 75% over baseline.
  • Developed an AI agent that generated production-ready ML repositories with batch and real-time inference workflows, REST APIs, and automated validation suites, reducing model onboarding time by 25%.
  • Expanded regression test coverage by 35% across 5 production models using AWS, Jenkins, and PyTest.

May 2025 – Aug. 2025

DATA ENGINEERING

Procter & Gamble

Data Engineering Intern

Cincinnati, OH

  • Built and deployed a full-stack sales forecasting platform with Plotly Dash, PostgreSQL, Docker, and Kubernetes to support promotion analytics and lift estimation.
  • Developed a baseline sales prediction algorithm, revealing $3M+ in profit optimization across key product lines.
  • Designed validation workflows in Databricks and SQL Server, reducing downstream anomalies by 33%.

May 2024 – Aug. 2024

SOFTWARE ENGINEERING

Procter & Gamble

Software Engineering Intern

Cincinnati, OH

  • Built a PyQt and Selenium form-automation tool that eliminated 750 hours of manual work annually.
  • Developed a reusable Java and Selenium testing framework that increased automated test coverage by 125%.
  • Built PySpark ETL pipelines in Databricks on AWS, cutting data delivery time by 80%.

Jan. 2026 – Present

COMPUTER GRAPHICS

The Ohio State University

Graduate Teaching Assistant

Columbus, OH

  • Support an upper-level graphics course covering Unity, C#, real-time rendering, and agent-based simulation.
  • Mentor 100+ students through office hours and review sessions while grading assignments, labs, and exams.
02 FEATURED QUESTS

Projects.

ARTIFACT 01
REPOSITORY INTELLIGENCE

RepoRover — Chat with a Codebase

A repository exploration and security-assessment tool that pairs conversational code questions with a navigable file tree and code viewer. The backend ingests GitHub repositories, chunks and embeds source code, and indexes it in Elasticsearch for retrieval-augmented responses.

1,000+ repository files indexed
PythonFlaskTypeScriptNext.jsReactLangChainOpenAI APIElasticsearchRAG
ARTIFACT 02
MOBILE DEVELOPMENT

Dissonance — Music Discovery & Reviews

An Android app for exploring the Discogs catalog and sharing album reviews. It combines Firebase-backed accounts and reviews with paginated discovery, ViewModel-based UI state, image loading, and resilient connectivity handling.

KotlinAndroidRetrofitFirebasePagingCoil
ARTIFACT 03
MODEL TRAINING & EVALUATION

LLM Fine-Tuning & Data Filtering

An experimental pipeline for testing how instruction-data filtering affects supervised fine-tuning. I compared length, lexical-diversity, deduplication, and combined filters on Llama 3.2 1B, with configurable training, LoRA adapter merging, and benchmark evaluation.

4 filtering strategies across 15 runs
PythonTinkerHugging FacePyTorchPEFTLoRAOLMES
ARTIFACT 04
COMPUTER VISION

Mathematical Symbol Recognition

A PyTorch image classifier that recognizes 16 handwritten mathematical symbols using a three-layer convolutional network. The project includes image preprocessing, training on 7,557 images, early stopping, and evaluation on a separate 1,010-image dataset.

16 symbol classes • 7,557 training images
PythonPyTorchCNNsComputer VisionJupyter
ARTIFACT 05
FULL-STACK DEVELOPMENT

TravelTabs — Shared Trip Expenses

A collaborative Rails app for managing group trips, recording shared expenses, allocating costs, and reviewing balances. My contributions focused on relational data modeling, authentication flows, email-based user lookup, validation, and trip and expense controller logic.

Ruby on RailsSQLiteDeviseTailwind CSSTurboStimulus
ARTIFACT 06
ROBOTICS & CONTROL

Airport Task Robot

A physical task robot built around closed-loop motion and environmental sensing. It uses PID control, wheel encoders, position correction, sensors, and servo-driven mechanisms to navigate and complete tasks reliably.

C++PID ControlEncodersSensorsServos
03 TRAINING & STUDY

Education.

The Ohio State University

Columbus, OH
GRADUATE STUDIES

M.S. Computer Science and Engineering

August 2025 – December 2026

GPA: 3.9 / 4.0

UNDERGRADUATE STUDIES

B.S. Computer Science and Engineering

August 2022 – December 2025

GPA: 3.9 / 4.0

04 TOOLS & DISCIPLINES

Technical Toolkit.

Languages

The foundations

PythonSQLJavaJavaScriptTypeScriptC#C++CKotlinRubyMATLAB

AI, ML & Data

Intelligence & insight

Machine LearningLLMsMulti-Agent SystemsPyTorchTensorFlowKerasHugging FacePySparkDatabricksTinkerPEFT / LoRALangChainRAGVector SearchHybrid SearchETLNumPypandasscikit-learn

Web, Mobile & Graphics

Interfaces & experiences

ReactNext.jsNode.jsExpressFlaskREST APIsPlotly DashPyQtRuby on RailsTailwind CSSAndroidRetrofitUnity

Databases & Services

Knowledge & connections

PostgreSQLSQLiteSQL ServerElasticsearchFirebase AuthenticationCloud FirestoreOpenAI API

Cloud, DevOps & Testing

Systems that endure

AWSDockerKubernetesJenkinsCI/CDPyTestSeleniumGitLinuxJupyterGradleBundler
THE NEXT EXPEDITION

Contact.

I'm always happy to connect with engineers, founders, recruiters, or anyone solving interesting software problems.

Get in touch