Startups

The ML startup ecosystem is thriving. More than $20B was invested in 550 AI startups in Q2 2021 alone. Fifty of those startups received $100M. In our community, we are tracking ML startups focused on infrastructure, DevOps, MLOps, platforms, hardware, and tools. To see a summary list of startups, click here. 

Abacus AI

Company Profile
  • Founded: 2019
  • HQ: San Francisco
  • No. of Employees: ~39
  • Raised: $40M
  • Founders: Bindu Reddy (CEO), Arvind Sundararajan (CTO), and Siddartha Naidu (VP)
  • Product. MLOps Platform
  • Value Prop: Build models at scale and ease
Background

Abacus AI was founded by an renown AI engineers. Their mission is to simplifies MLOps lifecycle. 

The startup helps companies build, train, deploy, and monitor ML and deep learning models with ease. Some areas it excels in are data pipelines, data wrangling, on-line feature store, model training, and monitoring drift. 

Activeloop

Company Profile
Background

Activeloop refers to itself as the Database for AI. The startup offers an open-source version of the product. 

It specialty lies in working with computer vision workflows that is able to rearrange data into NumPy-like arrays on the cloud on native deep learning frameworks. 

Big Squid

Company Profile
  • Founded: 2009
  • HQ: Salt Lake City, Utah
  • No. of Employees: ~26
  • Raised: $18.4M
  • Founders: Chris Knoch (CEO)
  • Product. AutoML
  • Value Prop: Develop and deploy machine learning models within the existing analytics stack without coding
Background

Big Squid an AutoML platform that brings machine learning and predictive analytics to the business decision-maker & front lines of a business.

Its products and solutions fulfill the varying needs of business leaders, data engineers, data scientists, BI developers, and many more professionals.

Comet

Company Profile
  • Founded: 2017
  • HQ: New York
  • No. of Employees: ~50
  • Raised: $69.8M
  • Founders: Gideon Mendels and Nimrod Lahav
  • Product. MLOps
  • Value Prop: Manage and optimize the ML lifecycle
Background

Comet is an MLOps platform that enables users to build ML models and compares experiments such as hyperparameters, predictions, system metrics, and more. 

In addition, models can be easily debugged, monitored, and alerts sent out when issues arise. 

Determined AI

Company Profile
  • Founded: 2017
  • HQ: San Francisco, California
  • No. of Employees: ~21
  • Raised: $13.6M
  • Founders: Evan Sparks (CEO), Ameet Talwalkar (Chief Scientist),  and Neil Conway (CTO)
  • Product. Open-source deep learning training platform
  • Value Prop: Determined is the fastest and easiest way to build deep learning models
Background

Determined is a reputed open-source deep learning training platform. It makes developing models easy, efficient, and fast.

Determined enables users to focus on building models by leveraging high-performance distributed training, GPU scheduling, state-of-art hyperparameter tuning, and model management in a single, fully-integrated environment.

Domino Data Lab

Company Profile
  • Founded: 2013
  • HQ: San Francisco
  • No. of Employees: ~289
  • Raised: $123M
  • Founders: Nick Elprin (CEO) and Chris Yang (CTO)
  • Product: MLOps platform 
  • Value Prop: Build, train, deploy, and manage models 
Background

Domino Data Labs provides an enterprise MLOPs platform .that is used by 20 of the Fortune 100 companies. The platform is used in the areas of medicine, crops, risk management, the auto industry, and few others. 

Domino works across the entire MLOps lifecycle and supports a wide array of frameworks and tools including TensorFlow, Spark, R Studio, MATLAB, etc. 

H2o.ai

Company Profile
  • Founded: 2012
  • HQ: Moutainview
  • No. of Employees: ~324
  • Raised: $246.5M
  • Founders: Cliff Click and Sri Satish Ambati
  • Product: AutoML platform 
  • Value Prop: Easy to use AutoML platform
Background

H20 is a unicorn that provides AutoML services via its cloud-based platform. The three platform themes are 1) make 2) operate and 3) innovate. 

  • Make: Build robust ML models
  • Operate: Performance monitoring
  • Innovate: Deliver solutions via AI AppStore  

Iguazio

Company Profile
  • Founded: 2014
  • HQ: Herzliya, Israel
  • No. of Employees: ~92
  • Raised: $72M
  • Founders: Asaf Somekh (CEO), Orit Nissan-Messing (VP R&D), Yaron Haviv (CTO), and Yaron Segev (CPO)
  • Product. MLOps Platform
  • Value Prop: Fully automated platform from data transformation to deployment
Background

Iguazio provides a data science platform that simplifies the MLOps lifecycle.  

The company provides tailored solutions to businesses operating in all industries, particularly financial services, retail, manufacturing, energy & utilities, healthcare, and smart mobility. 

Landing AI

Company Profile
  • Founded: 2017
  • HQ: Palo Alto
  • No. of Employees: ~74
  • Raised: $57M
  • Founders: Andrew Ng
  • Product. MLOps for Manufacturing
  • Value Prop: Helps manufacturers solve visual inspection problems
Background

Landing AI is one of the few startups with an MLOps platform focused on a specific industry, in this case manufacturing. 

Their product LandingLens is a comprehensive suite of products and features that enables users to create ML visual solutions. 

Latent AI

Company Profile
  • Founded: 2018
  • HQ: Menlo Park, California
  • No. of Employees: ~24
  • Raised: $22.5M
  • Founders: Sek Chai (CTO), Jags Kandasamy (CEO)
  • Product. MLOps for the edge
  • Value Prop: Compresses and optimizes neural nets at the edge 
Background

Latent AI, Inc. develops solutions that make the adaptive edge to transform AI processing. The company is well-funded by industry-leading investors with support from leading Fortune 500 clients.

Latent AI Efficient Inference Platform (LEIP) is a modular, fully integrated workflow that helps train, quantize, and deploy edge AI neural networks. 

Neptune AI

Company Profile
  • Founded: 2017
  • HQ: Warsaw, Poland
  • No. of Employees: ~42
  • Raised: $4.7M
  • Founders: Piotr Niedzwiedz (CEO)
  • Product. Metadata store for MLOps data
  • Value Prop: Manages metadata across experiments 
Background

Neptune AI lets you log, display, organize, compare and query your MLOps metadata. User can experiment tracking and model registry developed for research as well as production teams that run a lot of experiments. 

At present, more than 10K ML engineers and researchers manage their experiment and model metadata in Neptune. 

Neural Magic

Company Profile
  • Founded: 2017
  • HQ: Cambridge, MA
  • No. of Employees: ~28
  • Raised: $20M
  • Founders: Nir Shavit and Alexander Matveev
  • Product. Libraries that sparsify deep learning models for CPUs and GPUs
  • Value Prop: Open source libraries that optimize performance of models on CPUs
Background

Neural Magic unlocks the potential of machine learning by re-imagining deep learning. The company aims to sparsify deep learning models to reduce footprint and run on CPUs at GPU speeds.

Core components of their ML software architecture are Sparsify, SparseML, SparseZoo, DeepSparse Engine.

Neuro Inc

Company Profile
  • Founded: 2018
  • HQ: San Francisco
  • No. of Employees: ~16
  • Raised: $2M
  • Founders: Constantine Goltsev (CTO) and Maxim Prasolov (CMO)
  • Product. MLOps platform
  • Value Prop: Supporting the management of the infrastructure and processes for effective ML development at scale. 
Background

Neuro is a platform that supports the full end-to-end MLOps lifecycle for machine learning and deep learning models.  

The startup helps with data collection, model development, training, experiment tracking, deployment, and monitoring. 

OctoML

Company Profile
  • Founded: 2019
  • HQ: Seattle
  • No. of Employees: ~75
  • Raised: $132M
  • Founders: Jard Roesch, Jason Knight, Luis Ceze, Theirry Moreau, and Tianqi Chen
  • Product. MLOps platform
  • Value Prop: Deploy and manage ML models at scale 

OctoML is an MLOps platform that automates the deployment and ML models in the cloud. Supports most popular frameworks and a wide range of processors including CPUs, GPUs, and microcontrollers. 

The OctoML team developed the popular Apache TVM, an ML compiler framework for CPUs, GPUs, and other accelerators. 

Pecan.ai

Company Profile
  • Founded: 2016
  • HQ: Ramat Gan, Israel
  • No. of Employees: ~110
  • Raised: $50M
  • Founders: Zohar Bronfman (CEO) and Noam Brezis (CTO)  
  • Product. AutoML platform 
  • Value Prop: Deploy ML models at scale using any framework on Kubernetes
Background

Pecan.ai is an AutoML platform that simplifies feature engineering, feature selection, model training, algorithm optimization, monitoring, and predictive analysis. 

The startup works with different industries including demand forecasting, churn & retention, sales analytics, gaming, ecommerce, manufacturing, eLearning, and more.    

Pinecone

Company Profile
  • Founded: 2019
  • HQ: San Francisco
  • No. of Employees: ~23
  • Raised: $10M
  • Founders: Edo Liberty (CEO)
  • Product. Database for AI 
  • Value Prop: Vector database designed from the ground up for machine learning
Background

Pinecone is a fully managed vector database that enables vector search. The traditional database of organizing data into tables is not idea vector data represented by lines, pixels, points, text document, spatial, or the like. 

Pinecone enables users to search through billions of vectors in milliseconds. The founding team developed AI infrastructure at AWS and Yahoo. 

Roboflow

Company Profile
  • Founded: 2019
  • HQ: Des Moines, Iowa
  • No. of Employees: ~14
  • Raised: $22.2M
  • Founders: Brad Dwyer (CTO), Joseph Nelson (CEO)
  • Product. Computer vision models
  • Value Prop: Provide your software the incredible power to see objects in videos and images
Background

Roboflow is an MLOPs platform for computer vision. Users can build and deploy computer vision models with ease. Simply drag and drop datasets into their platform, and it will take it from there, including labeling. 

The platform serves all industries, including healthcare & medicine, manufacturing, agriculture, aerial & drone, infrared, and many more.

Run AI

Company Profile
  • Founded: 2018
  • HQ: Tel Aviv, Israel
  • No. of Employees: ~53
  • Raised: $43M
  • Founders: Omri Geller (CEO), Ronen Dar (CTO)
  • Product. Cloud-native compute orchestration for AI
  • Value Prop: Helps organizations manage GPU resource allocation and increase cluster utilization
Background

By virtualizing expensive hardware to pool, share, and distribute resources wisely, Run:AI lets enterprises perform on their AI efforts fast while keeping costs under control.

Its major platforms include Virtual Pool of GPU Machines, Kubernetes Scheduler, Gradient Accumulation, Machine Learning Pipelines, and more.

Seldon

Company Profile
  • Founded: 2014
  • HQ: London
  • No. of Employees: ~69
  • Raised: $13.7M
  • Founders: Alex Housley (CEO)
  • Product. MLOps platform 
  • Value Prop: Deploy ML models at scale using any framework on Kubernetes
Background

Seldon is a London-based startup that developed an MLOps platform that helps organizations deploy models at scale using any framework. It’s flagship product is Seldon Core, an open source platform that deploys ML jobs on Kubernetes at scale and leverages an extensive array of 3rd party products. 

Snorkel

Company Profile
  • Founded: 2019
  • HQ: Palo Alto
  • No. of Employees: ~92
  • Raised: $135M
  • Cofounders: Alex Ratner, Chris Re, Paroma Varma, Braden Hancock, and Henry Ehrenberg
  • Product. Programmatic labeling platform  
  • Value Prop: Labels data. And trains, improve performance, and deploys models
Background

Snorkel started off as a platform that programmatically labels datasets. Today, they’ve become more of a full-service platform that is able to build models, train, deploy, and monitor. The company was started by Stanford PhDs. Customers include Google, IBM, Intel, Microsoft, Uber, Accenture, and many other large enterprises. 

Google uses Snorkel to “replace 100k+ hand-annotated labels in critical ML pipelines.”  

Spell

Company Profile
  • Founded: 2017
  • HQ: New York
  • No. of Employees: ~30
  • Raised: $15M
  • Founders: Serkan Piantino (CEO) and Trey Lawrence (CTO)
  • Product. MLOps platform focused on deep learning
  • Market: NLP, computer vision, an speech recognition
Background

Spell is a NY based startup that has developed an MLOps platform focused on deep learning. It allows customers to do any number of experiments easily and cost-effectively in a fully automated manner. 

The platform runs on AWS, GCP, and Azure. And it is set up to use the right mix of CPUs and GPUs for workloads. In the AWS environment, the platform is designed to use lower-cost spot instances and although there might be disruptions in the use of these services, users will not be impacted since the system works around the dynamic nature of spot instances.

Verta

Company Profile
  • Founded: 2018
  • HQ: Palo Alto, California
  • No. of Employees: ~23
  • Raised: $10M
  • Founders: Manasi Vartak (CEO)
  • Product. MLOps platform
  • Value Prop: ML delivery and operations for high-velocity data science and ML teams
Background

The Verta MLOps platforms help data scientists and ML teams build and deploy ML models. The 4 steps of the platform are experimentation, register, deploy, and monitor. 

Users can set up their models using automated CI/CD workflows that work across the entire lifecycle from model training to deployment to governance.

Weights & Biases

Company Profile
  • Founded: 2017
  • HQ: San Francisco, California
  • No. of Employees: ~108
  • Raised: $65M
  • Founders: Chris Van Pelt (CVP), Lukas Biewald (CEO), Shawn Lewis (CTO)
  • Product. MLOps platform
  • Value Prop: Create better models faster and easier with experiment tracking, model management, and dataset versioning
Background

Weights & Biases provides a wide array of performance visualization tools for machine learning. 

The company helps businesses transform deep learning research projects into well-deployed software by supporting teams track, visualize and automate their models. 

WhyLabs

Company Profile
  • Founded: 2019
  • HQ: Seattle
  • No. of Employees: ~17
  • Raised: $65M
  • Founders: Alessya Visnjic, Andy Dang, Maria Karaivanova, and Sam Gracie
  • Product. AI Observability
  • Value Prop: Collects data and metrics from all parts of the MLOps lifecycle
Background

WhyLabs is an AI observability platform that collects performance data and metrics from all phases in the ML lifecycle. Thereafter, it provides users with actionable insight.

Engineers can monitor the quality of the data during training and production, then help pinpoint quality issues if any arise. The end-to-end observability platform detects data drift, stale models, and data quality changes. 

 

Scroll to Top