Sown To Grow
At Sown To Grow, I contributed across two tracks: building and iterating on public-facing website pages using React and TypeScript, and researching and improving NLP models in Python for classifying student reflection responses. The classification work involved understanding the nuances of student language and iterating on models to better surface meaningful patterns in how students reflect.
Design Process
Website Development
NLP Research & Model Iteration
Website Development
Developed and refined public-facing pages for the Sown To Grow marketing site using React and TypeScript, translating design into clean, maintainable code.
- Built pages in React with TypeScript, including the technology overview page
- Worked within an existing codebase to maintain visual consistency across the site
- Collaborated with the design team to implement responsive layouts accurately
NLP Research & Model Iteration
Researched and iterated on NLP models in Python for classifying student reflection responses — the core of Sown To Grow's insight engine.
- Analyzed student reflection datasets in Python to understand language patterns and edge cases
- Experimented with classification approaches and model architectures to improve label accuracy
- Evaluated model performance against labeled reflection data and iterated based on results
Outcomes
Shipped polished React/TypeScript pages supporting the product's public presence
Improved NLP classification accuracy for student reflection responses
Next Project