Michael Perkins

• Built ETL pipelines on Domo BI, processing 1M+ records/month to boost model accuracy by 45% and cut manual errors.
• Partnered with key stakeholders to define data needs and engineered self-service dashboards for informed budgeting.

• Designed and launched a React/Node.js app using ML models to evaluate game outcomes and deliver actionable insights.
• Identified market demand and built a platform serving 250 active users with innovative game strategies.
• Secured seed funding with persuasive VC proposals and a business plan outlining key value propositions.

• Automated processes with Python/SQL across hiring, capacity planning, onboarding, and finance, gaining 60% efficiency.
• Conducted data analysis to drive model improvements and optimize workflows for AI-driven decisions.
• Led digital rebranding (website and social media), increasing click-through rates by 20% and conversions by 15%.

GPA: 3.7

Relevant Coursework: Advanced Algorithms, Machine Learning, Software Engineering, Operating Systems

Capstone Project: Developed a ML model to detect move quality in real-time chess games.

Executive Editor at the Dartmouth Review

• Implemented a multi-stage SQL-to-NL pipeline using fine-tuned T5 transformers with PyTorch & Hugging Face.
• Improved BLEU scores through iterative dataset and hyperparameter optimization.
• Coordinated with teammates to align objectives and streamline deployment.

• Built Python/TensorFlow models achieving 95% accuracy in detecting chess anomalies.
• Mitigated imbalanced data issues with augmentation, reducing false positives by 20%.
• Integrated neural networks and real-time processing; coordinated with experts to refine detection criteria.