Platform Architect — AI and Cloud, Google
Vikas Mishra
I design AI agent systems and the platforms they run on at Google. My work sits where harness engineering, Google Cloud AI, and open-source AI meet, and I write and speak about what it takes to run agents in production.
Areas of Expertise
AI Agents & Harness Engineering
Agent orchestration, harness engineering, evals, and the loops and graphs that make agents converge in production
Open-Source AI
Open-weight models like Gemma and Llama, vLLM serving, LoRA fine-tuning, and open tooling wired into enterprise stacks
AI/ML Architecture
LLM pipelines, model serving, MLOps, TPU/GPU optimization for enterprise AI systems
Cloud Solutions
Google Cloud Platform, multi-cloud architecture, cloud migration, and infrastructure optimization
Distributed Systems
Microservices, Kubernetes, scalability patterns, and high-availability architectures
Engineering Leadership
Team building, technical mentoring, and engineering strategy for AI/ML teams
16+
Years of Experience
5x
GCP Professional Certified
30+
Conference Talks
60+
Engineers Mentored
Speaking & Advisory
I speak about AI agents, harness engineering, and production AI
Keynotes and technical deep-dives for conferences and engineering organizations: agent orchestration, Google Cloud AI, open-source AI, and what it actually takes to run agents in production. 30+ talks delivered globally.
Recent Writing
Agent Orchestration Is a Graph Problem, Not a Prompt Problem
Why linear agent chains waste most of their wall-clock, how to find the parallelism hiding in your workflow, and what actually breaks when you fan out to a fleet.
Your Agent Loop Doesn't Know When to Stop
An agent can always produce another answer. Whether the loop converges or burns money depends on four things, and none of them live inside the model.
The Forward Deployed Engineer Is Not a Solutions Architect in a Hoodie
The fastest-growing role in AI, reverse-engineered from the inside: what a Forward Deployed Engineer actually does, what the interview really tests, and why the model was always the easy part.
Google's ADK Is a Runtime, Not a Graph: Notes From Eleven Agents
Eleven agents in, the framework choice that mattered wasn't ergonomics or graph syntax. It was whether the runtime had opinions about events, state, and transport. ADK does. Here is what that buys you in production.
Built to Be Cited: An Engineer's Guide to AEO and GEO in 2026
An implementation-grade walkthrough of AI-search optimization in 2026, centered on Google AI Mode and Gemini, with the schema graph, llms.txt work, IndexNow wiring, and the debugging sessions that shipped the result — drawn from rebuilding my own site.
Certifications & Badges
Blockchain-Based Verification for Fair Play in Fantasy Sports Using Merkle Trees
The AI Valley Podcast
Conversations on AI, cloud architecture, and the future of technology.
Projects
AI Coding Playbook
An open-source, MIT-licensed book and reference kit for agentic coding with Claude Code and Gemini CLI — 21 chapters, runnable slash commands, skills, hook scripts, and end-to-end workflows.
EPUB2PDF Print Perfect
A Python tool that converts EPUB eBooks into beautiful, print-ready PDFs with preserved typography, embedded images, auto-generated tables of contents, and clean chapter breaks.
Explain Error
A CLI tool that transforms cryptic error messages and stack traces into clear, actionable explanations using AI-powered analysis.
Repo Inspector
A CLI tool that uses AI to analyze codebases and explain what they do, how they're structured, and how data flows through them.