We designed Rockset to enable organizations of all sizes, from global enterprises like those in the Fortune 500 to small startups with just a few members, to build and scale highly efficient search and AI capabilities seamlessly within the cloud. Today, we’re proud to mark a significant achievement in our pursuit of revolutionizing search and analytics for the era of artificial intelligence. Here’s the improved text:
The round is led by Icon Ventures, joined by new investors Glynn Capital, 4 Rivers, and K5 International, alongside our existing backers at Sequoia and Greylock. With this funding, our total capital now stands at $105 million, allowing us to move forward with confidence into the next phase of our growth strategy.
Classes discovered from @scale deployments
I spearheaded the growth of Facebook’s online information infrastructure from its inception in 2007, when it boasted 30-40 million monthly active users, to a peak of 1.5 billion MAUs by 2015. Initially, Facebook’s pioneering Newsfeed operated in batch mode, relying on primary statistical models for ranking and refreshing content only once every 24 hours. Throughout my time, Fb’s engagement skyrocketed as Newsfeed turned the world’s hottest advice engine powered by superior AI & ML algorithms and a strong distributed search and analytics backend. Our team collaborated on developing seamless transitions that span from enabling the Like button functionality to delivering personalized advertisements, and ultimately, thwarting spam and other malicious activities. The success we’ve achieved is a direct result of the robust infrastructure we carefully built. Our CTO, Dhruba Borthakur, pioneered RocksDB, while our chief architect, Tudor Bosman, led the Unicorn initiative, which underpins all search functionality at Facebook; he also developed infrastructure for the Facebook AI Research Lab. Meanwhile, I designed and scaled TAO, the underlying technology that fuels Facebook’s social graph capabilities. Witnessing firsthand the profound impact that accessing top-notch data has on personal growth and development.
Following the groundbreaking introduction of ChatGPT, countless businesses embarked on exploring the capabilities of artificial intelligence (AI), marking a significant turning point in their technological trajectories. As companies scale their successful ideas into production, it is essential that they consider three critical factors:
- Streaming-first architectures are a critical foundation for the AI era. By reimagining traditional dating apps with eco-friendly features, users can now connect with like-minded individuals while minimizing their digital footprint. For instance, incorporating real-time alerts about who’s currently online or within a specific geographic radius can foster meaningful interactions and reduce the need for constant swiping. The airline’s innovative chatbot, designed to address travellers’ pressing concerns amidst ever-changing weather conditions and flight schedules.
- AI advancements are unfolding with steady momentum. If your team is solely focused on managing pipelines and infrastructure instead of continuously iterating on functionality, keeping pace with emerging trends will become increasingly challenging?
- While AI functions can be incredibly powerful, they can also quickly become expensive and resource-intensive. The ability to efficiently scale applications in the cloud will enable businesses to unlock the potential of AI.
What we consider
We believe that innovative search and AI applications in the cloud should prioritize environmental sustainability and scalability.
Engineers worldwide should be able to rapidly develop powerful data applications with ease. Developers should be able to build innovative apps without being restricted by proprietary APIs and arcane programming languages that require an extensive investment of time and effort. Building these apps must seem straightforward, akin to crafting a well-structured SQL query that yields the desired results with ease.
Trendy information apps should operate on information in real-time. One of the most effective apps excel at providing a comprehensive dashboard for businesses and their customers, rather than being a simple reflection of what’s behind them?
We believe that innovative information apps should inherently prioritize environmental sustainability. To ensure optimal performance, sources should automatically scale up to meet growing demands without requiring additional management, and also scale down regularly to prevent unnecessary expenses and optimize resource utilization. The genuine benefits of cloud computing are fully unlocked when you opt for a “pay-per-use” model rather than reserving “energy” upfront.
What we stand for
We relentlessly pursue efficiency, leaving no stone unexamined.
- We developed RocksDB, a highly acclaimed and widely used high-performance storage engine.
- We pioneered a revolutionary converged index storage format designed to optimize computing resources, facilitating efficient information indexing and retrieval within compute environments.
- Developed from scratch in C++, our high-performing SQL engine achieves response times of sub-10 millisecond speeds.
We dwell in real-time.
- Our team has successfully developed a cutting-edge real-time indexing engine that boasts a remarkable 4-fold environmental advantage over industry-leading platforms like Elasticsearch.
- Built on top of RocksDB, our indexing engine enables efficient data mutability, including upserts and deletes, without incurring standard performance penalties for such operations.
We exist to empower builders.
- A unified catalog system? Enable seamless querying of disparate data types by indexing JSON, vector embeddings, geospatial, and time-series information within a unified database framework in real-time. Utilize query capabilities across ANN indexes for vector embeddings, as well as JSON and geospatial “metadata” fields with maximum efficiency.
- If you’re familiar with SQL, you’re likely well-versed in structuring queries to extract insights from your data. Similarly, when using Rockset, you’ll find that its syntax is designed to be SQL-like, making it easy to leverage your existing knowledge to get started.
We are consumed by a relentless pursuit of efficiency within the realm of cloud computing.
- We developed a groundbreaking, singular database that pioneered the concept of compute-separation, providing unparalleled isolation between computations. Create a Virtual Event for Seamless Streaming of Information Consumption. Host a brand-new virtual event seamlessly within your app, effortlessly connecting attendees worldwide. Eliminate redundant competition by scaling each entity uniquely to optimize its unique strengths. No need to worry again about performance slumps resulting from ingestion surges or query floods.
- We developed an exceptionally efficient auto-scaling hot storage tier leveraging NVMe SSDs. Innovative performance fusion enables lightning-fast input/output operations, effortlessly handling even the most resource-intensive projects.
- By leveraging auto-scaling compute and auto-scaling storage, you only need to pay for the resources you actually use. Don’t break the bank with unnecessary cloud clusters.
AI-native search and analytics database
Traditional first-generation indexing technologies like Elasticsearch were initially designed for on-premise environments, predating the widespread adoption of artificial intelligence and the need for real-time updates.
As AI advancements continue to surpass expectations, Large Language Models (LLMs) and generative AI applications are unlocking previously inaccessible information trapped within unstructured data sources. Superior AI models transform text, images, audio, and video into vector representations, requiring powerful tools to store, index, and query these embeddings to build a cutting-edge AI application.
While AI applications may require similarity search and nearest neighbor search functionalities, traditional k-nearest neighbor (kNN)-based solutions can be computationally expensive and inefficient for these tasks. Rockset leverages the power of FAISS, enabling the creation of high-performance ANN indexes that can be seamlessly updated in real-time and efficiently queried alongside diverse metadata fields, thereby simplifying the development of sophisticated search and AI applications.
Within unique parameters of 1?
Was a significant challenge for our small team to overcome? The constant drain on productivity was severely compromising our ability to effectively develop the intelligence of our advice engine, hindering its capacity to keep pace with our growing ambitions. We need to add a brand-new customer signature to our analytics pipeline in order to effectively capture and track their data. By leveraging our existing serving infrastructure, we can efficiently transmit data through Confluent-hosted topics, utilizing both Kafka streams and ksqlDB. This enables the subsequent normalization and aggregation of information as needed. The selected Elasticsearch index must be manually curated to effectively process and store the relevant data. Only under exceptional circumstances could one reasonably doubt the accuracy of this data. The entire process took weeks.
Maintaining our existing inquiries required a tremendous amount of effort. As our data evolves constantly, we consistently integrate fresh updates into existing databases. Requiring a tedious and labor-intensive update to the associated Elasticsearch index with every adjustment necessitated a considerable amount of time and effort. As each was built or updated, it became essential to meticulously review the entire information pipeline to verify that newly introduced components did not introduce inefficiencies, propagate inaccuracies, and the like.
Testimonies aligning with diverse opinions highlight that various stakeholders concur that effective ML and AI adoption necessitates a focus on developing AI-driven applications rather than optimising the foundational infrastructure to support cost scalability. Rockset is an AI-native search and analytics database designed with these specific goals in mind.
We intend to leverage the additional funding to expand our reach into new territories, intensify our marketing initiatives and drive forward our technological advancements in this domain. Join us on this transformative journey as we reimagine the boundaries of search and AI capabilities by delving into the innovative world of Rockset, empowering you to begin your exploration today. I’m excited to see what’s next in store.
