■ NewsInsight from the intersection of AI and business
Product updates, research, customer stories, and perspectives from the team building AI at S&P Global.
NewsArchive
Building on moving ground: Life as an MLOps engineer at Kensho
Over the past two years as an MLOps engineer at Kensho, I’ve come to appreciate both the challenges and the immense value of operating in an environment where implementation patterns, tools, and the surrounding ecosystem are constantly changing.
How Kensho built a multi-agent framework with LangGraph to solve trusted financial data retrieval
Discover how Kensho, S&P Global’s AI innovation engine, leveraged LangGraph to create its Grounding framework–a unified agentic access layer solving fragmented financial data retrieval at enterprise scale.
Enhanced document extraction for generative AI use cases
Introducing Kenverters, a set of open source Python developer tools for fast and easy use of Kensho Extract in pipelines for Retrieval Augmented Generation (RAG), Large Language Models (LLMs), table extraction, and more!
An Introduction to mitigating toxicity in LLMs - Pt. 2
In our last post, we introduced toxicity as a challenge when building LLMs and various methodologies to mitigate it. In this post, we’re going to focus on utilizing one of those methodologies — auxiliary tools — and assessing some tools that are available to the public.
An introduction to mitigating toxicity in LLMs
A vital aspect of “productionalizing” Large Language Models (LLMs) is to ensure a safe user experience and promote alignment with societal standards. LLMs excel at generating language and answering user-input questions, even some of the hardest ones. However, when given a malicious prompt, LLMs may also try to answer that with an unsafe response.
Postgres and OpenSearch: Mature data storage technologies and high-performance vector search engines
Postgres and OpenSearch: Mature Data Storage Technologies and High-Performance Vector Search Engines
Learnings from the lab: Querying S&P Global’s tabular data using LLMs
Understanding complex data structures is a major challenge for Generative AI use cases. Kensho is designing specialized LLM-Ready APIs to tackle this problem.
Distributed training with Kubernetes
Neural networks are more relevant than ever with the rise of GenAI, in particular large language models, and at Kensho we’ve been employing them in a wide range of applications. However, training these complex models can be computationally intensive and expensive. In collaboration with our ML Research team, Kensho’s infrastructure team tackled this challenge using a multi-node approach that could be easily integrated into our existing infrastructure.
How human-reviewed machine-readable transcripts supercharge audio data
Machine-Readable Human-Reviewed (MRHR) transcripts merge modern technological capabilities with human reassurance to produce a transcription that is machine-readable and human-reviewed.
Automating application deployments at Kensho
Kensho’s AI product offerings encompass about 20 applications, including core services as internal services like a debugging dashboard and a customer authentication tool. As our AI products promote efficiency for our customers, we are also always trying to automate internal workflows, such as deployments, as well. A six-member SRE — Site Reliability Engineering — team is responsible for working with development teams to deploy these services to production clusters, collectively known as Hydra.
Kensho NLP hackathon: Exploring new challenges with natural language processing
Kensho’s NLP team organized the company’s first ever NLP-specific hackathon that brought together engineers, designers and product managers to explore and develop experimental products using NLP techniques. The hackathon was a great team-building event for connecting, collaborating, showcasing skills and creating new solutions.
Word Error Rate primer
Word Error Rate (WER) is a commonly used metric to judge the accuracy of Automatic Speech Recognition (ASR) systems. It is the ratio of errors in the ASR’s transcription to the total number of words spoken. Although there are other metrics to measure how well an ASR system is performing, WER is a relatively simple but actionable metric that allows us to compare ASR systems to one another, and evaluate an individual ASR system’s accuracy over time.
What I’ve learned after 1.5 years as a Junior Software Infrastructure Engineer
Software infrastructure engineers maintain the infrastructure for the software of a company. We create, manage, and maintain the tools and services that the other software engineers use. After only a year and a half in the role, I still have a lot to learn, but I’ve also learned more in that time than I could have imagined.
■ Get startedTalk to our team about what’s possible
At Kensho, you’ll work at the intersection of trusted data, advanced AI, and real-world decision-making. We hire talented people and give them autonomy and support to develop cutting-edge solutions that solve complex business problems.