How Anthropic Engineer Gurmeet Saran Builds Trustworthy AI Systems
- Gurmeet Saran joined Anthropic as a member of technical staff in San Francisco in 2025, taking on the engineering challenge of building infrastructure to monitor how large language...
- Saran grew up in Kolkata and graduated from high school in 1999 before earning a bachelor’s degree in electronics and telecommunication engineering from the University of Kalyani in...
- To solve conflicting website metrics, Saran created a single source of truth model that defined metrics such as daily active users.
Gurmeet Saran joined Anthropic as a member of technical staff in San Francisco in 2025, taking on the engineering challenge of building infrastructure to monitor how large language models produce outputs and keep private user data secure, spectrum.ieee.org reported. Saran develops systems to understand the mechanics behind chatbot behavior to ensure that the artificial intelligence architecture delivers reliable and trustworthy results.
Building Data Reliability at Yahoo and Intuit
Saran grew up in Kolkata and graduated from high school in 1999 before earning a bachelor’s degree in electronics and telecommunication engineering from the University of Kalyani in 2004. He later completed a master’s degree in software engineering from the Birla Institute of Technology and Science in Pilani in 2010. His professional career began in 2007 as a technical lead for the data team at Yahoo in Bengaluru, India, where he addressed metric discrepancies across different company divisions.
To solve conflicting website metrics, Saran created a single source of truth model that defined metrics such as daily active users. His team subsequently built Yahoo! Digits, an internal analytics tool deployed across company products. He moved to Intuit as a staff engineer in Mountain View, California, in 2010, where he helped construct data analytics infrastructure and real-time tax fraud detection features that complied with Internal Revenue Service policies.
Scaling Metric Governance at Meta and Wealthfront
In 2016, Saran joined Meta as a lead software engineer in Menlo Park, California, following public scrutiny over overstated video view metrics. He engineered Metric360, a standardized system designed to govern all video view metrics across the platform. By his departure in 2021, the system was fully integrated into the site architecture. He then worked as a principal software engineer at Wealthfront in Palo Alto, California, before moving to Anthropic.
Reflecting on his career across multiple tech firms, Saran stated that privacy and governance must be embedded in software architectures from the beginning.
As I learned earlier in my career, privacy can’t be a layer you bolt on—it has to be part of the architecture from the start.
Gurmeet Saran
Developing Observability Tools for Large Language Models
At Anthropic, Saran focuses on the observability challenges presented by large language models, comparing the work to medical monitoring equipment.
Picture a physician observing a patient’s vital metrics and choosing a course of treatment relying upon those figures. What if one or a few of the medical devices used were slightly miscalibrated? Should the vast majority of measurements be accurate, this remains a grave issue because critical choices are made near the boundaries.
Gurmeet Saran
He notes that tracking conversational agent actions while protecting sensitive user data remains a demanding engineering hurdle. In March 2025, Saran joined IEEE as a senior member to contribute to peer review and technical publishing, advocating for the establishment of formal engineering standards for artificial intelligence infrastructure.