In stealth · 2026

Engineered for tokens.

We're building an operator-grade inference cloud — quietly, for now.

// Built by operators. More soon.

Get in touch → We're hiring →
What we're building

Under wraps — for now.

We'll share more as we come out of stealth. Investors and operators who want the full picture early can reach us directly.

soon
Approach
soon
Architecture
soon
Traction
The team

Built by operators who've run GPU + network infrastructure at hyperscale.

The credible part we'll happily show. The rest comes later.

PR
Prithvi Rai
Co-founder

Founder of Borneo (AI/data security & governance, acquired by Atlassian). 20+ years at Uber, Meta and Yahoo building infrastructure platforms at scale.

AA
Amit Agarwal
Co-founder

20+ years building infrastructure and platforms at Google, WhatsApp and Cisco. IIT Kharagpur & Ohio State alumnus.

SL
S. L.
Co-founder · in stealth

25+ years operating cloud GPU and web-scale infrastructure across leading systems and hyperscaler organizations. Name on request.

SL
S. L.
Co-founder · in stealth

20+ years building and operating network + GPU infrastructure at scale, including across NeoClouds. Name on request.

GoogleMetaCiscoUberCrusoeDropboxAtlassian
Careers · Founding team

Help us build the inference factory.

We're hiring founding engineers to build a cloud from the silicon up. Small team, deep ownership, real metal. These are the profiles we're looking for — if one sounds like you (or close), reach out.

Apply / say hi → ← Back to home
Founding Engineer · Inference

Inference & kernels

  • CUDA kernel authoring and tuning
  • Cross-platform compilation across heterogeneous silicon
  • Inference optimization — prefill/decode, batching, KV-cache
Founding Engineer · Network

GPU network at scale

  • Design / deploy / operate IB & RoCE clusters
  • Nvidia & AMD GPU fabrics at 20k–50k scale
  • Topology, congestion and collective-comm performance
Founding Engineer · Datacenter Ops

Operations & build

  • Rack and floor layouts, structured cabling
  • CDU / PDU design and power distribution
  • Bring-up, commissioning and day-2 operations
Founding Engineer · Systems

Drivers & low-level systems

  • Kernel drivers, hardware & NIC firmware
  • NCCL / RCCL / CUDA libraries
  • End-to-end performance optimization on NCCL

Don't see your exact role?

We're early. If you've operated GPU + network infrastructure at scale and want to build the token layer on top, we want to talk.

careers@tokenforge.ai →