Transformers, LLM Tuning, RAG Expertise

We: a consultancy that builds internet things for money, and sometimes for other reasons. We are currently in need of a clever programmer with good taste and plenty of experience, and an interest in the state of the art. Our team is good people, we love the things we work on, do our best to keep low stress and good humor, and only ask that you truly give a damn.

Right now, we are convinced a major change is that search is being replaced. We are looking for somebody who has been following these trends for awhile, understands transformer architecture and the attention mechanism down to the QKV matrices, and has experience applying open-source systems.

Some of the problems on our horizon include big data optimizations with large models, intelligent / next gen GPU scale management, retrieval / cache augmented generation, SOTA 3D processing systems, AR / VR rendering, and autonomous vehicle construction. We like the fun stuff, and we hope you do too!

Our Idea Candidate

• Is capable of finding recent papers, and rapidly testing them.
• Has the ability to dive deep into the code to optimize those systems. This generally requiress a deep comfort with pytorch.
• Basic experience with fine-tuning existing models
• Experience with openAI, Anthropic, and other API
• Is interested in understanding how things actually work, rather than plug and pray
  
  

Nice To Haves

• Comfortable with the docker ecosystem, so that differnet projects with different dependencies can easily be strung together
• Experienced with FastAPI
• Interested in other categories of machine learning, like 3D / vision
• Comfortable dealing with PostgreSQL