Our Work

A small sample of our work:

• 

  How We Helped Google Introduce Tensorflow Probability to the World

  Tensorflow Probability added a lot of then-much-desired probabilistic programming features to the already versatile Tensorflow 1.x. Here is the announcement blog post written for some of our past work with the Google Tensorflow Team on expanding Tensorflow Probability.

  Tensorflow Blog (external link, see Matthew McAteer's name in the author list)
• 

  Before BitTensor reached $420M+ market cap, we helped them go from 0 to 1

  Out of concern for the increasingly siloed large AI models, the Bittensor team was looking to devise distributed training protocols that would let such neural networks be trained across organizations. Here's how we helped Bittensor go from 0 to 1, which eventually went on to become a $420M+ market cap project.

  by Matthew McAteer
• 

  Refining Liquid Measurement for Lab Automation Using AI With Monomer Bio

  Monomer bio is revolutionizing lab automation. We collaborated with Monomer Bio on a liquid measurement system relying on ultrasonics and computer vision.

  by Matthew McAteer
• 

  Foresight Institute is allocating $1.2M for Novel AI Safety Research

  AI Safety is similar to nuclear safety: The future of the human race might literally depend on it. Unfortunately AI Safety research tends to be underfunded compared to AI capabilities research. We're working with the Foresight Institute to help them decide on the most promising recipients of more than a million dollars in AI safety research funding.

  by Matthew McAteer
• 

  Weni deploying LLaMa-2 models for 50% cheaper and with 6X better latency

  Weni is using Large Language models based on LLaMa-2 for retrieval augmented generation. Unlike most companies, Weni is now able to get these results 6X faster and 50% cheaper than before. Here's how a high-level overview of how we achieved this.

  by Matthew McAteer
• 

  Building the Future of Hiring with YouTeam (YC W18)

  YouTeam is building the futue of contractor qualification. We worked with them to build a chatbot that could reliably make business-critical decisions from customer requests using state-of-the-art retrieval-augmented-generation (RAG).

  by Matthew McAteer
• 

  Building an AI Drug Discovery pipeline so HairDAO can cure hair loss

  HairDAO is a decentralized research & development group devising novel treatments for hair loss in a transparent way. Here's how we helped them build an AI Drug Discovery pipeline to narrow down a library of 48,000 molecules down to just 5 promising candidates.

  by Matthew McAteer
• 

  11 companies, 11 industries, 11 ML services

  5cube Labs has worked with more than 50 companies across many industries. We've helped companies with domains ranging from computer vision to natural language understanding, to ML Ops, to making sense of tabular data. Here are eleven examples of the diverse services we provide.

  by Matthew McAteer
• 

  Working With O'Reilly Media to Put Our Best Practices Into a Book

  We've worked with O'Reilly media to release a book on Practicing Trustworthy Machine Learning. We cover everything from privacy-preserving ML, to interpetaility, to robustness, to brand new topics like homomorphic encryption, synthetic data, and causal inference. Available now!

  Book Website (external link)
• 

  A Preview of Our 50-Step Process for up to 2,500X OpenAI API Cost Reduction

  Providers of APIs for powerful Large Language models (LLMs) like OpenAI's GPT-4 can seem like magic, BUT if you're not careful they can burn a hole in your wallet. Here's a story of how we devised a system to help companies identify cost-savings areas in their LLM usage that could save them hundreds of thousands of dollars at scale.

  by Matthew McAteer