Abstract: Today, commercial deployments of MLOps at a scale are going through an evolutionary change across almost all verticals and use cases. Notably, deployment and product at a scale are seeing a whole slew of commercialization issues arise as “edge cases” become more significant at scales of thousands, millions or even billions. As a result, “traditional pre-deployment training” is no longer sufficient to ensure quality and reliability standards in production, especially in edge devices. Personalization at the edge - cards, phones, cameras, etc. need to be continuously trained with custom & specific data relevant to their individual usage or environment to be safe and accurate. For example, the neural networks that govern the way an autonomous vehicle drives in New York City is not the same in Kansas City or Mexico City, and surely not in Yellowstone National Park. The explosions of data streams from billions of “sensors” in real & virtual worlds need to be constantly processed for retraining models and data preparation is getting extremely complex as use-cases strive to replicate normal human behavior.
The selection of a DataOps partner to navigate this scale, fragmentation, volume and complexity is undergoing major metamorphosis, too. Selection criteria based on basic workforce efficiency and cost metrics is not sufficient. In fact, they could be counter-productive to your goals of robust, production-ready ML applications. Key KPIs that are fast emerging as critical for almost all ML applications such as data preparation and annotation (DataOps) is no longer a “one way street” process feeding the MLOps cycle, but is intricately fused to it at multiple stages. The talent ask is now beyond someone who can draw polygons around objects in an image and they also need to be subject matter experts with significant experience in a vertical or use-case. The usage of a “crowd” is no longer the optimal solution for many applications. The technical ability (technique) needed for these complex DataOps projects can only be groomed through long-time employment, experience, and comprehensive training. Technology’s role in DataOps has become more important than ever before. A robust workforce, project, task, and data management platform is one of the most critical pieces to cater to the scale. Usage of tooling to improve productivity, efficiency, accuracy as well as automation are also must-haves in an efficient, scalable DataOps platform. For example, Edge case analysis is now a critical part of scaling in production. An edge case in a pilot deployment will surely be unacceptably common in a commercial fleet. This requires tools and talent to work synergistically, with a deep understanding of the client’s application and training methodology. iMerit ensures continued success for your MLOps application, development, evaluation and tracking metrics across all 3 criteria above is very critical. Only a few DataOps companies have managed to make the transition and provide a robust integrated end-to-end solution. A still smaller fraction can keep pace with the scale, performance and automation roadmaps of this fast-changing industry, and iMerit is at the helm as one of the industry leaders.
Bio: Rajsekhar Aikat is iMerit’s newly appointed Chief Technology & Product Officer. Rajsekhar joins iMerit from Qualcomm, where he was Senior Director of Product Management. He has over 18 years of technical & product experience across multiple verticals, including automotive, IOT, robotics and telecom. Before Qualcomm, he was the Director of Product at Brain Corporation, where he was responsible for scaling BrainOSTM, an autonomous mobile robot platform & ecosystem, as well as overseeing the development and commercialization of commercial cleaning and delivery robots globally.