Michelle van

About Michelle van

I design and deliver data and machine learning systems that transform raw data into scalable, intelligent solutions. My work bridges data engineering, applied ML, and solution design, enabling automation, analytics, and measurable product impact.In fast-paced, high-growth environments, I’ve operated across multiple disciplines—from data pipelines and applied ML to analytics and solution architecture—connecting technical execution with business strategy to deliver systems that scale.At Velvet FS, I’ve built cloud-native data pipelines, performance attribution frameworks, and LLM-powered systems that streamline investor reporting and insights. I’ve scaled data workflows, enhanced reliability, and built solutions that empower teams with timely, trustworthy insights.Previously, I helped Policygenius integrate NLP models (BERT, GPT) into production pipelines—cutting analysis time from three weeks to one day—and at Perlego, I modernized the backend to a serverless AWS architecture, improving latency and platform performance.I’m passionate about building applied data and ML solutions that are scalable, interpretable, and genuinely useful. I thrive in roles where I can connect engineering, machine learning, and data strategy to create systems that drive real-world outcomes.Core Skills: Python, SQL, Airflow, BigQuery, AWS, dbt, PyTorch, scikit-learn, Hugging Face, NLP, LLMs, Data Modeling, Tableau

Experience

1. Ml Engineer Data Scientist

   Velvet

   Sep 2023 — Present · San Francisco, CA, US

   Built document processing systems that transform unstructured PDFs into structured datasets, automating financial reporting and analytics workflows •Designed cloud-native data pipelines and performance frameworks powering portfolio analytics and investor insights •Developed LLM-powered tools for intelligent data extraction and search, reducing manual effort and turnaround time •Led requirements scoping, architecture design, and implementation, aligning data and ML solutions with business objectives and user needs

Education

• University of Oxford

  Master of Engineering - MEng, Materials Science

• University of Oxford

  Bachelor's and Masters of Engineering degree, Materials Science

• Duke University

  Master of Science - MS, Data Science