Build
Better
AI with Veris

A human-in-the-loop approach to synthetic data generation for every ML team in the world.

Get Started
A Peek Behind-The-Scenes

How Does It Work?

Our technology works on a simple three-step loop: Generate, Feedback, Repeat

Step 1
Generate

Use a seed dataset to guide the training of a generative model, which outputs new, synthetic samples to augment your original dataset

Step 2
Feedback

Distribute the new samples to a set crowdworkers, verify that the output appears real, and remove any unrealistic or unwanted data.

Step 3
Repeat

Use crowdworker evaluations to refine the data and models - continue this proccess until sufficient samples are realistic.

Our Use Cases

Veris provides a rapid and accurate way for autonomous vehicle manufacturers to produce, analyse, and verify vital sensory data.

Autonomous Vehicles

Our Thesis: Using Humans to Improve Synthetic Data

Veris promises a future of high-quality data for all. We plan to accomplish this through three key components: "smart" synthetic data generation, human-in-the-loop crowdworker networks, and an iterative feedback process.

Why Use

Synthetic Data

Our "smart" synthetic data complements your existing dataset to give you exactly what you need to train the best model you can.

The Need For

Humans

As ML engineers, we know just how important high-quality data is. That's why every sample we produce is carefully analyzed and vetted by a real person.

Ensuring Better Data Through

Feedback

Crowdworker feedback allows us to improve our generative models with each iteration, making our data generation of higher quality than incumbents.

Meet Our Team

Veris is powered by an experienced team of researchers and engineers.

Aditya Rai

Avner Lipszyc

Ajil Jalal

Sriram Vishwanath

Sandeep Chinchali