reppertj / triton-inference-server Goto Github PK

A machine learning engineer with a background in applied statistics 📊 and the philosophy of mathematics, I like building practical tools using machine learning models and thinking about ML infrastructure ☁️, measuring and monitoring model performance ⏱, and ETL pipelines. I work at Ought, where we're building Elicit, an AI-powered research assistant.

Check out some of my projects:

• Earworm 🎸, a sound-based search engine for music licensed for royalty-free commercial use. It uses PyTorch, React, FastAPI, Postgres, TypeScript, and TensorflowJS and can be scaled in production using Docker swarm mode.
• Image Captioning 🖼, a framework for experimenting with image captioning architectures in PyTorch and PyTorch Lightning.
• Philosophical Graphiti, a webapp for exploring and visualizing relationships among philosophical topics. It uses Django, Postgres, NetworkX, and Vega.

I have a PhD in Philosophy, with a dissertation defending a classical phenomenological approach to the philosophy of mathematics. Talk to me about Edmund Husserl, formal semantics, or impredicativity!

Before and during graduate school, I helped support international development evaluations for companies including Social Impact, DevTech, and the QED Group. Ask me about using experimental methods to learn about development effectiveness and making stats-heavy reports clear and readable to subject-matter experts and management.

Read more at justinreppert.com or reach out to talk via LinkedIn or email at justin [at] justinreppert [dot] com.