Ingero Team

An open-source research project focused on kernel-level GPU observability and tracing CPU-GPU interactions using eBPF

Joined on Mar 31, 2026 https://ingero.io

Cover image for Agent + MCP + eBPF: 10,869 CUDA Kernel Events, Now Queryable

Ingero Team

Apr 21

Agent + MCP + eBPF: 10,869 CUDA Kernel Events, Now Queryable

#gpu #ebpf #mcp #observability

5 min read

Cover image for 11-Second Time to First Token on a Healthy vLLM Server

Ingero Team

Apr 21

11-Second Time to First Token on a Healthy vLLM Server

#vllm #observability #ebpf #mcp

5 min read

Ingero Team

Apr 16

What Happens When an AI Agent Gets Kernel-Level GPU Traces

#gpu #ebpf #observability #gpuobservability

5 min read

Ingero Team

Apr 16

MCP as Observability Interface: Connecting AI Agents to Kernel Tracepoints

#ebpf #mcp #gpuobservability

5 min read

Ingero Team

Apr 13

One Query, Four GPUs: Tracing a Distributed Training Stall Across Nodes

#gpu #ebpf #distributedcomputing

7 min read

Ingero Team

Apr 8

CUDA Graphs: The 8-Year Overnight Success and the Observability Gap

#cuda #gpu #ebpf #ai

9 min read

Ingero Team

Apr 1

124x Slower: What PyTorch DataLoader Actually Does at the Kernel Level

#pytorch #gpu #python #cuda

5 min read

Ingero Team

Mar 31

Tracing a 13x PyTorch Slowdown to a Hidden NumPy Synchronization

#pytorch #cuda #python #gpu

4 min read

Forem

Ingero Team

Badges

Writing Debut

Skills/Languages

Currently hacking on

Agent + MCP + eBPF: 10,869 CUDA Kernel Events, Now Queryable

11-Second Time to First Token on a Healthy vLLM Server

What Happens When an AI Agent Gets Kernel-Level GPU Traces

MCP as Observability Interface: Connecting AI Agents to Kernel Tracepoints

One Query, Four GPUs: Tracing a Distributed Training Stall Across Nodes

CUDA Graphs: The 8-Year Overnight Success and the Observability Gap

124x Slower: What PyTorch DataLoader Actually Does at the Kernel Level

Tracing a 13x PyTorch Slowdown to a Hidden NumPy Synchronization