I build software, data systems, and machine learning projects that solve real-world problems.
My background started in business and customer success, but I kept finding myself automating processes, analyzing data, building tools, and asking "why are we still doing this manually?" Eventually I stopped treating engineering as a hobby and started treating it as a craft.
Today I'm focused on data analytics, machine learning, cloud infrastructure, and building systems that people can actually use.
- 📊 Data analytics, SQL, and business intelligence
- 🤖 Machine learning with PyTorch
- 🏗️ Data pipelines and ETL workflows
- ☁️ Cloud infrastructure and self-hosted services
- 🧠 Retrieval-Augmented Generation (RAG) systems
- 🚀 Building projects that move from idea → production
Python • SQL • PostgreSQL • Power Query • Pandas • NumPy • Databricks (learning)
PyTorch • Scikit-Learn • MLflow • Computer Vision • Transfer Learning
FastAPI • Node.js • Docker • Linux • AWS • Cloudflare • Terraform
React • Next.js • TypeScript • Tailwind CSS
Computer vision project that classifies bird species from images using transfer learning and PyTorch.
- Trained on the CUB-200-2011 dataset
- Supports EfficientNet-B0 and ResNet50 architectures
- YAML-driven training configuration
- MLflow experiment tracking
- Foundation for future inference APIs and deployment
Tech: PyTorch • MLflow • Python • Computer Vision
A self-hosted Retrieval-Augmented Generation system that answers questions about my projects, technical writeups, architecture notes, and portfolio content.
Visitors will be able to ask:
- "How does BirdBrain work?"
- "What ML experience does David have?"
- "Which projects use Python?"
- "Explain the architecture of Sentinel."
Planned Stack: PostgreSQL + pgvector • FastAPI • Open Source LLMs • Docker • Cloudflare
Image processing platform built on AWS infrastructure.
Designed to explore scalable image delivery without relying on large third-party SaaS platforms.
- Signed URL workflows
- CloudFront distribution
- Infrastructure as Code
- Automated deployment pipelines
Tech: AWS • Terraform • Python
A social discovery platform for finding restaurants and places through recommendations from people you trust.
- Interactive mapping experience
- User-generated recommendations
- Modern React architecture
- Backend powered by Supabase
Tech: React • TypeScript • Supabase • Mapbox
My personal playground for learning infrastructure, Linux, networking, containers, and deployment patterns.
Current experiments include:
- Dockerized services
- Reverse proxy configurations
- Self-hosted applications
- Monitoring and observability
- CI/CD workflows
- AI and ML deployment experiments
Because production skills are usually learned after breaking things first.
I'm currently deepening my skills in:
- Dimensional modeling
- Star schema design
- ETL/ELT architecture
- Data warehousing concepts
- Databricks
- Distributed data processing
Many of my projects increasingly blend software engineering, analytics, and machine learning into a single workflow.
A project isn't finished because it works once.
It's finished when:
- It can be run again without heroics
- Failures are observable and diagnosable
- The next developer can understand it
- Changes don't require a rewrite
- Documentation exists before it's desperately needed
If future-me can't understand it six months later, it wasn't done.
💻 https://github.com/djmartin2019
I'm always interested in discussing data engineering, analytics, machine learning, cloud infrastructure, and building useful software.