Accelerate AI Inference With Intelligent Context Memory
Reuse KV Cache across requests, GPUs, and infrastructure to reduce TTFT, increase token throughput and GPU efficiency.
Transform the Economics of Distributed Inference at Scale
Our Vision
Who We Serve
The Problem
Make memory a first-class infrastructure layer for AI inference
As AI moves toward longer context, reasoning, RAG, and agentic workloads, efficiently managing and reusing context becomes critical. TensorMem's intelligent memory infrastructure layer orchestrates working memory (KV Cache) across memory tiers and across distributed inference—making it faster, more scalable, and more efficient.
The Problem
Who We Serve
The Problem
AI Inference Recomputes Too Much
As context windows grow and workloads become more agentic, inference repeatedly processes the same or overlapping context. Valuable computed state is trapped in limited GPU memory or discarded, forcing expensive recomputation. Reusing this state reduces TTFT, increases token throughput, improves GPU efficiency, and lowers inference cost at scale.
Who We Serve
Who We Serve
Who We Serve
Built for AI Inference at Scale
We serve AI clouds and neo-clouds, inference platforms, serverless inference providers, enterprises, and AI infrastructure providers. Wherever long-context, RAG, reasoning, or agentic workloads demand higher throughput and better GPU efficiency, TensorMem helps make inference faster and more economical.
Completely software-defined. No new hardware!
Founding Team & Advisors
Anand Kekre (Co-founder)
Arvind Pande (Co-founder)
Arvind Pande (Co-founder)
IIT Bombay graduate, serial entrepreneur with two successful startup exits and 67 US patents. Former Chief Architect and CTO at Veritas and Arctera. Deep expertise in Distributed Software-defined Storage, Infrastructure, Cyber Resilience, and cloud-scale enterprise solutions.
Arvind Pande (Co-founder)
Arvind Pande (Co-founder)
Arvind Pande (Co-founder)
IIT Kharagpur graduate, 15 US patents, and founding engineer of MapR Technologies - a pioneering distributed file system platform for large-scale analytics. Deep expertise in high-performance data infrastructure and hyper-scalable enterprise solutions.
Fred van den Bosch (Advisor)
Srinivasan "Sesh" Seshadri (Advisor)
Srinivasan "Sesh" Seshadri (Advisor)
A private investor, business consultant, and experienced technology executive. His career spans the architecture and design of complex hardware and software systems, as well as leadership roles including CEO, CTO, and Executive Vice President of Engineering. He previously served on the Board of Directors of VERITAS Software and currently
A private investor, business consultant, and experienced technology executive. His career spans the architecture and design of complex hardware and software systems, as well as leadership roles including CEO, CTO, and Executive Vice President of Engineering. He previously served on the Board of Directors of VERITAS Software and currently serves as a Director of Varonis Systems, Inc., as well as a board member and advisor to several private companies.
Srinivasan "Sesh" Seshadri (Advisor)
Srinivasan "Sesh" Seshadri (Advisor)
Srinivasan "Sesh" Seshadri (Advisor)
Srinivasan “Sesh” Seshadri is a three-time founder (Zettata, Kosmix, and Strand Life Sciences) with successful exits and a veteran of database and distributed systems, with 30+ years across academia, startups, and large companies. Most recently, he was Chief Innovation Officer at Aerospike. He has also held senior roles at Google and Bell
Srinivasan “Sesh” Seshadri is a three-time founder (Zettata, Kosmix, and Strand Life Sciences) with successful exits and a veteran of database and distributed systems, with 30+ years across academia, startups, and large companies. Most recently, he was Chief Innovation Officer at Aerospike. He has also held senior roles at Google and Bell Labs and was a computer science faculty member at IIT Bombay. Sesh holds a PhD from UW-Madison and a B.Tech from IIT Madras, and has 50+ publications and 15 issued patents in databases and data systems.
Investors
We are backed by "Venture Guides", and experienced technology leaders and angel investors from Google, VERITAS, VMWare, Netflix, NetApp and many other leading infrastructure companies.
About TensorMem
TensorMem Inc. is a Delaware C-Corp building a software-defined context-memory orchestration platform for high-performance inference, powered by an AI-native distributed caching and storage layer.
As inference becomes the dominant driver of AI economics, TensorMem addresses the growing challenge of managing KV Cache and data movement across the memory–storage hierarchy.
Built on deep expertise in distributed systems, enterprise storage, databases and resiliency, TensorMem provides a foundational layer for efficient, scalable AI inference that works across models, inference servings and hardware platforms.
