John Simonsen

MLOps Stack Q&A Bot | End-to-end RAG Application

• Developed a Retrieval-Augmented Generation (RAG) application designed to answer questions about MLOps tools (ZenML, MLflow, Docker, etc.) by utilizing official documentation.
• Engineered a data ingestion pipeline involving web scraping, embedding generation (using Ollama), and storage in a ChromaDB vector database.
• Built and containerized a RESTful API using FastAPI and Docker, and integrated it with a front-end web interface.
• Deployed the containerized application on a local Kubernetes cluster for development and testing.
• Authored a complete blueprint for cloud deployment, including Terraform configuration for AWS (EKS, ECR, S3) and a CI/CD pipeline using GitHub Actions.

Flight Delay Predictor | MLOps, CI/CD & Model Deployment

• Developed and deployed a machine learning model for predicting flight delays using Python, FastAPI, Docker, and MLflow.
• Utilized MLflow for model tracking, versioning, and deployment, showcasing experience with a leading MLOps platform.
• Implemented a CI/CD pipeline using GitLab to automate the testing (Pytest) and deployment process.

Customer Sentiment Analysis | NLP & LSTM Networks

• Developed and trained an LSTM neural network model using PyTorch to classify customer sentiment from product and service reviews.
• Implemented text data preparation pipeline including cleaning, tokenization (NLTK), lemmatization, and padding.
• Employed techniques like embedding layers, dropout regularization, and early stopping to mitigate overfitting.

Alignerr (Remote) & DataAnnotation (Remote)

• Contribute to specialized AI model evaluation projects, assessing capabilities in code generation, technical problem-solving, and multi-language instruction adherence to refine model performance.
• Perform comparative analysis on AI model variations by evaluating side-by-side outputs to guide development and enhance user interaction quality.
• Provide detailed feedback on AI-generated code and technical outputs, identifying errors and logical flaws to improve model reasoning, accuracy, and robustness.

Northern Utah Academy for Math, Engineering, and Science (Layton, UT)

• System Administration: Administered and maintained a 28-workstation Ubuntu lab, managing updates, user provisioning, and troubleshooting.
• Automation & Tool Development: Developed automation scripts (Bash, Python) and CI/CD pipelines (GitHub Actions) to streamline classroom operations, including automated grading and system imaging.
• Mentorship: Led student teams in the CyberPatriot competition for five years, consistently achieving top rankings (1st-3rd) in state competitions.

Western Governors University

Utah State University

Awards

• Eagle Scout Award with Bronze Palm
• Winner, Northrup Grumman Coding Challenge, Hill Air Force Base