AI FOR GOOD
SCHLOSS BOSS AI
Matt Schlosser | Frisco, TX
About
A software developer used to specialize in one tool like a violinist. Pick your instrument, practice until your fingers bleed, master the solo. That era is ending. With the growing power of AI, one engineer can step into the conductor's chair and lead the entire orchestra at a blistering pace. That is what I do. With 300+ hours of AI-accelerated development and a Georgia Tech MS in Analytics, I deliver production ML models, full ETL pipelines, and complete SaaS applications. Solo.
Before founding Schloss Boss AI, I taught AP Statistics and BC Calculus for over a decade, chairing the math department and helping more than 1,000 students earn college credit. When I ran out of mountains to climb in education, I went back to school, earning an MS in Analytics from Georgia Tech (4.0 GPA) and getting early hands-on experience with LLMs by building NLP models from scratch during my practicum internship. That foundation launched my engineering career: end-to-end data pipelines, ML prediction models, and executive data visualizations at a cybersecurity SaaS.
Experience
- AI consulting, data engineering, and SaaS development using Claude Code and agentic workflows to deliver full projects as a solo engineer
- Built and deployed a complete SaaS application (FastAPI, React, PostgreSQL) with fuzzy employer deduplication and automated job ingestion pipeline
- Ship production ML models, ETL pipelines, and full-stack applications end-to-end for clients
- Engineered customer churn prediction model (0.92 AUC, Gradient Boosting) deployed in a Tableau dashboard used daily by Customer Success for over a year
- Tested 5 model types (Logistic Regression, Random Forest, Gradient Boosting, XGBoost, LightGBM) with rigorous cross-validation and class weighting
- Implemented sentiment analysis pipeline using Snowflake Cortex for support ticket classification and summarization
- Built end-to-end ML pipelines: Snowflake data assembly, Python/sklearn modeling, Tableau delivery with prescriptive actions per account
- Built executive dashboards in Tableau including "Support Performance" and "Channel Partner Performance"
- Led ML/AI projects using Snowflake Cortex and custom Python solutions
- Designed data pipelines with dbt and Fivetran ensuring data integrity for mission-critical KPIs
- Built NLP models from scratch using XLM-RoBERTa to classify instances of bullying in unstructured text conversations
- Applied transformer and neural network techniques using Python (PyTorch) and Hugging Face on large-scale text data
- Chaired mathematics department of 20+ teachers, overseeing curriculum development and strategic planning
- Helped 1,000+ AP Calculus BC and AP Statistics students earn college credit
Projects
GitHub →Anti-algorithm social platform built solo with Claude Code. Next.js, Firebase, real-time feeds — no recommendation engine, no engagement tricks.
Real-time volleyball stat tracker. Built in a weekend using Claude Code agentic workflows.
“Matt was a valuable asset on our data analytics & BI engineering team. His knowledge of machine learning modeling, AI, SQL, data warehousing & data engineering, as well as data visualization made him highly competent. He was also skilled on the soft skills side — gathering customer requirements, presenting to stakeholders of all levels, and working with the rest of the team.”
Writing
The shift from coding specialist to AI orchestra conductor. Why your coding background is the prerequisite, not the casualty, of AI-accelerated development.
Skills, plugins, agentic workflows, and MCP, exploring how a CLI-based AI coding assistant changed the way I build software.
Thirteen years of teaching notes, inference frameworks, and scoring rubric patterns encoded into a single Claude Code skill for basic statistical inference.
End-to-end ML pipeline at a PE-backed SaaS company, going from zero predictive capability to a deployed Gradient Boosting model with 0.92 AUC.
Education
Top-5 nationally ranked interdisciplinary program spanning Computing, Engineering, and Business. The program is called Analytics, not Data Science, because the craft goes beyond precision. Analytics is the art of turning data into decisions that move people and organizations.
Every model was built from the ground up. No black boxes. We learned the mathematics, wrote the code, and understood why it worked before we ever touched a library. That foundation proved instrumental during my practicum, where I implemented neural networks and attention mechanisms to build NLP models from scratch.
Graduate-level coursework in ML/AI, statistical modeling, NLP, data pipelines, visualization, and optimization.
Georgia Tech Project Portfolio
Built NLP classification models from scratch using XLM-RoBERTa and PyTorch to detect bullying in unstructured text data from a children’s messaging platform.
Compared 13 ML models (XGBoost, Random Forest, LASSO, SVM, neural networks) with cross-validation to predict NBA player salaries from 47 advanced statistics including PER, RAPTOR WAR, and on/off efficiency sourced from Cleaning the Glass, FiveThirtyEight, and Basketball Reference.
Interactive data visualization tool for exploring university data across multiple dimensions including academics, demographics, and outcomes.
Contact
Interested in working together? Let's connect.