David Nelson
I’m a software engineer based in Brazil. I work on web products, data-heavy systems, and developer workflows. Over the years I’ve worked as both an individual contributor and an engineering leader, and I tend to enjoy the places where technical depth, product judgment, and team direction meet.
Currently
I’m working at Dexory on software that turns warehouse robot scans into inventory visibility, exceptions, and operational workflows.
Things I keep returning to
- AI-assisted software development
- not as magic, but as a change in how software gets written.
- Data-heavy applications
- modeling information well, making queries fast, and keeping product logic understandable as systems grow.
- Product-minded engineering
- building software around real workflows, feedback loops, and the decisions people need to make.
- Home automation and personal infrastructure
- small systems that are useful because they keep working.
- Technical leadership
- especially the parts close to product judgment, architecture, and team direction.
Code
A few recent public projects and older experiments live on GitHub.
- sentiment-analysis-imdb — An LSTM experiment on film reviews.
- winston — A small constraint-satisfaction implementation for Ruby.
- notaflow — tools for manual Brazilian NFS-e generation from commercial invoices.
- nfse-js — TypeScript tooling for Brazil's national NFS-e standard.
- ivce — a local-first CLI for contractor invoices from YAML files.
- jsonresume-theme-colophon — a warm, text-first JSON Resume theme for HTML resumes and PDFs.
- GitHub archive — older public code, experiments, and smaller repos.
Publications
Research work on machine learning and financial time series.
- Stock market’s price movement prediction with LSTM neural networks — IJCNN, 2017.
- Using LSTM and Technical Indicators to predict price movements — KDMiLe, 2016.