Data Science

Aporia Emerges with $5M to Ensure AI Integrity

Aporia allows data scientists to quickly create their own custom monitors for machine learning models

Aporia, the production observability SaaS platform for machine learning, left stealth today and announced the launch of the first customizable monitoring platform for Machine Learning models with full support for private and public clouds. The company also revealed $5M in seed funding from Vertex Ventures and TLV Partners and is already being used by multi-billion dollar companies. Aporia enables data scientists to quickly and easily build their own monitors, so they can keep track of their Machine Learning models’ performance, ensure data integrity and provide responsible AI.

Companies across the globe are investing upwards of $50 billion dollars annually on AI adoption but a lack of observability and little ability to quickly detect issues in ML  models as they run in production is undermining their investment.

The only way to know if a complex system is working as expected is to monitor it. Nevertheless, Machine Learning models are challenging to track both technically and conceptually as they rely on real world data to accurately make predictions about the future. Machine Learning models can work perfectly in the experimentation phase, but they start to drift over time due to engineering changes in databases and APIs – or more often due to constantly-changing data and new events occurring in the real world. Something as routine as a company expanding into a new market, or as dramatic as the Covid-19 pandemic can have significant implications on the performance of a model.

When Machine Learning models perform poorly, customers and businesses suffer the consequences. Predictions based on the wrong data are often flawed, resulting in unintended outcomes such as a customer being shown the wrong recommendations or an accidental refusal of a loan application. This, in turn, can cause lost revenue or expose companies to discrimination and unfairness claims. Without proper monitoring, it can take months before a company even notices that its models have stopped making accurate predictions.

“AI needs guardrails”, says Liran Hason, CEO of Aporia. “Companies need to have confidence in their Machine Learning models, and the only way to get there is by robust monitoring to ensure they’re doing what they’re supposed to do.”

With Aporia, data scientists can create bespoke monitors for their Machine Learning models in just a couple of clicks, and set alerts of different severity to be sent to email or sources like Slack. Aporia’s monitors are extremely flexible, allowing data science teams to watch the right things for their own unique models and business cases.

Aporia can be installed with a few lines of code and monitors asynchronously, handling workloads of billions of daily predictions with no impact on latency. The user interface is full-featured, clean and simple, making it easy to create, maintain and modify monitors. Once Aporia’s platform reveals an issue, data scientists can often quickly track down the cause of the problem and decide how to address it, whether via a logic change, a bug fix or retraining the ML model when necessary.

Concerns regarding data security and regulations make many companies apprehensive about adopting public cloud monitoring tools. Alongside its public cloud deployment, Aporia offers an innovative “managed on-prem” solution, giving peace of mind to large companies and corporations with high data privacy and security requirements.

Liran Hason, the founding CEO of Aporia, is a veteran of the IDF’s elite 81 intelligence unit. He was one of the first employees of Adallom (acquired by Microsoft), where he led the ML production architecture, serving millions of users. Before starting Aporia, Hason was part of Vertex Ventures’ investment team, and was involved in over 30 investments including Axonius, and others.

“Companies are struggling to keep watch of their AI in the ways that matter for their specific Machine Learning model and use case,” said Hason. “Aporia makes monitoring simple, fast and secure, bringing engineering and DevOps best-practices into the new field of MLOps and ensuring that data science teams can keep their models performing accurately and fairly.”

Emanuel Timor, General Partner at Vertex Ventures says “AI adoption is soaring and requires a proper technological stack to handle the new challenges that come with it. Aporia is a vital part in the new MLOps stack, filling a critical gap in production readiness of AI.”

Rona Segev, Founding Partner at  TLV Partners: “Monitoring production workloads is a well-established software engineering practice, and it’s past time for machine learning to be monitored at the same level. Aporia’s team has strong production-engineering experience, which makes their solution stand out as simple, secure and robust.”

Related posts

Adapdix addresses data scientist shortage with a $5/hour program

PR Newswire

Mona Introduces Free, Self-Service Monitoring for GPT Applications

PR Newswire

Domino Data Lab leads in Two Independent AI-Focused Market Studies

PR Newswire