Alexander Small
i build cool stuff... (like this website)
About Me
01Welcome.
My name is Alexander, but all my friends call me Zander. I'm a Software Engineer specializing in Machine Learning and AI technologies. I hold a B.S. & M.S. in Software Engineering (Dual Degree) from St. Mary's University, completed from August 2021 to May 2025.
I have extensive experience developing production LLMs, building training pipelines, and creating scalable ML solutions. Currently based in New York, NY, I work on document classification systems and ML infrastructure at scale.
When I'm not building AI systems, you'll find me deep in the world of anime and all things weeby! This pic was actually taken while I was drifting JDM cars in the Japanese mountains (yes, really). If you're into anime too, I'd love to know your top 5 - it's always great to connect with fellow weebs in the tech industry!
My Skills
02Technical Skills
Languages & Libraries: Python, SQL, TypeScript, JavaScript, React, Pandas, NumPy, PostgreSQL, Regex, Scikit-learn
ML, Cloud & Infrastructure: AWS (SageMaker, S3, Textract), OpenAI API, LangChain, MLflow, Kubernetes, Docker, Snowflake, Datadog, LLMs/RAG, CI/CD
Notable Projects
HIPAA-Compliant GPT Reporting Tool
Built a local Python GUI integrated with the GPT API for a children's shelter, coordinating with OpenAI to ensure HIPAA-compliant rest API endpoints. Decreased report generation time by 95%, allowing staff to shift focus from manual documentation to direct care. View on GitHub
Shingo Quest - Japanese Road Sign Learning Platform
Built a full-stack gamified learning application using React, TypeScript, and Vite with Supabase backend, implementing Row Level Security policies, real-time data synchronization, and custom database triggers for user progress tracking. Developed comprehensive authentication system with XP-based progression, achievement badges, and streak counters to enhance user engagement and learning retention for Japanese road sign education. View on GitHub
My Work
03Data Scientist @ Coalition Inc. – New York, NY (June 2025 – Present)
Architected an org-wide Document Classification Service on AWS (SageMaker, S3) to automate claims processing for the Manual Submissions team, replacing a rigid legacy model with a modular system capable of identifying loss runs, financial statements, and signatures.
Achieved 99% accuracy (up from 79%) in document categorization by implementing Chain-of-Thought (CoT) prompting and designing rigorous evaluation pipelines using MLflow and Python.
Reduced token consumption by 40% and lowered overall inference costs by refactoring output logic (removing JSON brackets) and identifying cost-efficiencies in migrating from GPT-3.5 to newer GPT-4 models.
Deployed a unified API endpoint integrating Regex brand classifiers and LLMs, enabling the broader engineering organization to access scalable document intelligence tools.
Technical Founder @ Mavn AI – New York, NY (May 2024 – Jun 2025)
Selected to interview for Y Combinator (S24 & F24), placing the startup in the top 5% of applicants based on technical merit and traction.
Engineered end-to-end data pipelines on AWS SageMaker and Kubernetes, enabling the secure ingestion, preprocessing, and fine-tuning of datasets for production LLM deployment.
Built and deployed our full-stack annotation platform using React and TypeScript, integrating LangChain agents to automate quality assurance and reduce manual verification time.
Scaled data production operations to support 150+ concurrent annotators, optimizing PostgreSQL database performance to ensure data consistency across distributed engineering teams.
Delivered production-critical training datasets to industry partners (e.g., V7 Labs, DataCurve), directly supporting model benchmarking and evaluation cycles.
Software Engineer, Human Data @ Scale AI – San Francisco, CA (May 2023 – May 2024)
Engineered multi-stage QA pipelines for LLM training data, designing consensus algorithms and automated conflict resolution logic that improved dataset accuracy from 91% to 98%.
Developed automated validation frameworks using Python, Pandas, and Scikit-learn, implementing statistical checks to detect anomalies and rank annotator performance at scale.
Managed high-throughput data storage in PostgreSQL, supporting Kubernetes-deployed model training workflows processing thousands of daily tasks.
Software Engineer @ Ignite Local – Seattle, WA (Jul 2021 - May 2023)
Engineered a proprietary CMS in Python, deploying a PostgreSQL-backed architecture on AWS to centralize management for hundreds of client properties.
Scaled web production capabilities, automating the deployment of 300+ responsive websites built with React and TypeScript while ensuring cross-browser compatibility.
Software Engineer Intern @ Ignite Local – Seattle, WA (May 2021 – Jul 2021)
Contributed to early-stage web development and component design in React, leading to a return offer for a full-time software engineering role.
Contact Me
04You can contact me at alexanderpaulsmall@gmail.com with any questions or concerns (or to squad up in Marvel Rivals)
Career updates and more information about my qualifications. Feel free to message me if you have any questions about my qualifications or to schedule an interview.