The generative way to develop ML solutions
Skilled workers in data science and machine learning (ML) R&D
Is spent annually on wages for these workers
The time needed for R&D to support custom ML-based products
The current spend on ML R&D is not enough to solve the slow time-to-market for ML solutions. Even more challenging, there is a shortage of skilled ML engineers and data scientists. We believe that the solution begins in the process. Let's dig deeper…
Why is the R&D process so slow today?
ML engineers and data scientists must perform many experiments before arriving at an ideal model for a given task.
modlee to the rescue!
How it works
Use modlee model suggestions within your python scripts or with our CLI tool to bypass unnecessary ML experimentation.
Access any modlee branch for the groups you're connected to.
Incorporate the modlee package in your existing PyTorch scripts to document new experiments.
Store results and assets privately in your local environment.
Sign up for an API key.
Set up modlee to privately share your experiments with groups of co-workers and peers or the entire ML community.
Your choice, your control.
We train modlee on ML experiments shared with the community.
We routinely and privately fine tune modlee to group experiments.
We are releasing our beta in stages
Stage 1: ML experiment documentation
Stage 2: Community model suggestions
Stage 3: Private group model suggestions
Connect with us
reach out to firstname.lastname@example.org
Here is how we see modlee ...
modlee is the generative way to share ML knowledge. Our vision is to improve the way organizations develop machine learning by transforming the process to be collective, automated, and simple. There’s a race for generative ai and it’s already started. The global AI and ML market is projected to reach $1.6 trillion by 2030. We believe the future of generative ai is niche and specialized and that modlee can revolutionize ML R&D as new tools are essential to support the needs of this growing AI and ML market.
There are more than 10 million ML experiments in R&D today yet given the limited tools these experiments are often slow, redundant and tedious for researchers. Discovering a model that satisfies a task can even take months. Generative ai can learn from and leverage these efforts to provide a more effective and faster way to share and extend ML knowledge.
modlee is building a generative ai that will suggest the best model for any ML task, dataset and target requirements, with one-shot accuracy. modlee is a low code solution, agnostic to any data set, can easily integrate into standard ML code bases and workflows, will bypass tedious and redundant ML experimentation, and can learn securely from within your organization and from the greater ML community.
modlee is not a wrapper around ChatGPT, nor are we an iterative AutoML or ML Ops platform. We are however an essential R&D ML tool, your go-to coworker in ways, always working to accelerate the application of ML knowledge. We want to help researchers be even more effective innovators enabling each to have more time and a greater focus on breaking new ground.
Our team has deep AI and ML expertise, a successful track record building great companies, and passion about this space.
Tons of ML researchers, including our own, are excited for early access to modlee's beta!
"I love applied deep learning research, but performing extensive model architecture experimentation to fit the constraints of a specific problem is tedious. The experimentation that the ML community does as a whole seems so wasteful and under utilized given the overlap of ML knowledge within different industries. After having spent years in ML R&D, I knew there must be a better way to share ML knowledge more effectively and directly, and am so excited to share modlee with the world."
Brad Magnetta: modlee Founder, CEO & CTO
How does modlee use my data?
modlee analyzes your data to help determine the best model suggestion for your problem, but never copies or stores your data outside of your local coding environment.
Is modlee an AutoML process?
Traditional AutoML processes are iterative, whereas modlee is generative. The iterative AutoML process requires training several models and is thus slow and may not outperform human intuition. modlee learns model suggestions from community experimentation and documentation. modlee's model suggestions directly use your dataset and target specifications to provide one-shot model suggestion and training cycles for ML R&D.
Is modlee a wrapper around ChatGPT?
modlee is not simply a wrapper around ChatGPT. We train custom models on actual machine learning experiments to suggest the best models for your dataset and target requirements such as accuracy, model size, and model speed.
Is modlee an MLOps platform?
modlee is a package that can be added to any python coding environment. We work with existing ML frameworks (TensorFlow, Keras, PyTorch) as well as MLOps framworks (MLFlow, Vertex AI, etc).