Guest: Nima Negahban (LinkedIn)
Company: Kinetica (Twitter)
Show: Let’s Talk

Many of our readers may not know but Arlington (Virginia) is nothing short of ‘Silicon Valley’ or the East Coast with many tech start-ups emerging in this region, mostly serving the evergreen federal government, especially organizations like Defense Advanced Research Projects Agency (DARPA), DoD and so on.

Being based out of the DMV region ourselves, we have the privilege of talking to many of these companies. We will soon be starting a series of “TFiR Pioneers: Companies To Watch Out For” interviews focussed on DMV-based start-ups.

The idea of the series came up during my interview with Nima Negahban, Co-Founder and CEO of Kinetica, which has emerged as a leading Arlington (Virginia)-based, real-time data company.

The company was started in 2010 as part of a DoD research project to consume hundreds of different real-time data feeds, and then be able to give a query capability to analysts to data scientists, to developers to quickly be able to deploy stuff into the field.

The company not only closely works with government agencies like United States Postal Service (USPS) and DARPA but also caters to the private sector. I will sit down with Nima again to dive deeper into the story of the company at some point, but this interview focussed on how Kinetica has started to use generative AI. One of the generative AI use cases gaining traction is utilizing it to help enterprises to analyze real-time data.

In this episode of TFiR: Let’s Talk, Negahban discusses the evolution of data towards real-time data and how this is being used by enterprises for a competitive advantage. He goes on to discuss the key trends of generative AI and how Kinetica is helping enterprises leverage it for analyzing real-time data.

Highlights of this video interview:

  • Negahban talks about his background working on a DoD project and the challenges they faced with real-time data at a time when there were still legacy warehouses. He talks about how this led to the creation of Kinetica in 2010.
  • Kinetica was the result of flipping the idea of data being a scarce resource on its head, enabling data to continuously stream and to be able to write any query you want without data engineering, leveraging all of this abundant compute in a distributed way.
  • Kinetica has a large DoD customer base, but also includes customers from the financial and telecom sectors, helping them do advanced analytics that utilize the real-time data with historical datasets and to be able to query it without any limitation.
  • Negahban explains the importance of real-time data saying that it is we are really just seeing the start of it in the modern enterprise. He discusses how it enables the enterprise to understand what is going on in the business in real time and how this can help give them the competitive advantage over other players.
  • Negahban talks about the progress that has been made in the data space moving from static data to real-time data. He explains how Kinetica fits into the picture helping put these database capabilities into a single pane of glass so decision makers can better understand what is going on in the business.
  • Real-time data is playing an integral role in vector search and generation, something that Kinetica has been focusing on. Negahban discusses how the company is making it easy to do powerful brute force GPU vectors and vector similarity search. He talks about how GPUs are built for being able to do massive processing without having predefined and pre planned queries and indexes.
  • Negahban tells us about how knowledge graphs are being used for their ability to find more entity correlations. He talks about how knowledge graph and vector search together can provide a more powerful way to deal with structured data to power enhanced LLM workflows.
  • Generative AI is a hot topic right now particularly in two areas for Kinetica: powering vector search and having a natural language to SQL capability baked into the product. Negahban discusses how the generative AI revolution has acted as a catalyst in these areas.
  • Negahban talks about the key trends he is seeing with generative AI saying that full-featured vector search capability is now widely available across all databases. He talks about the evolution of this to being able to scale the vector search with data latency and how this is where Kinetica are seeing their advantage.
  • Although there was originally a lot of hype around ChatGPT, now that the dust is settling people are realizing there is a limit to how much they can do. However, Negahban feels there is still much you can do with it if it is used correctly and orchestrated correctly.
  • It is important for organizations to find a balance between staying with trusted, reliable technologies and embracing emerging technologies. Negahban discusses how Kinetica makes it easy for people to do their own exploration and data science in Kinetica. With other LLM products and services coming out, Kinetica is also looking for ways to offer support for a wide variety of generative LLM options.

This summary was written by Emily Nicholls.

You may also like