I lead applied AI engineering teams and enterprise-scale cloud architecture teams. The work spans modernization programs, reference architectures, distributed systems and platform engineering.

Agents are one strand of that. The one I build in public is Loki Mode, the engine behind Autonomi. Everything I write, ship and think about lives at asklokesh.com.

• Applied AI engineering: getting AI and agent systems out of demos and into production.
• Enterprise cloud architecture: reference architectures other teams can build on.
• Modernization: moving large, long-lived systems forward without stopping them.
• Distributed systems: the part I first studied in grad school and still care about most.
• Platform engineering: platforms that other engineers actually want to use.

It started in 2008 with an old Red Hat server in a college lab. A professor let me install RHEL on a dusty machine in the corner, and I learned more in two weeks than I had in a semester. I still remember the first time I SSHed in from across the hostel and felt like a hacker in a movie. I wrote about that, and the vi struggle, on the blog.

After a B.Tech in Electronics and Communication, I worked as a Linux L3 engineer running datacenter operations. Then came an MS in Computer Science at Texas A&M, where I was a research assistant working on distributed systems.

From there I moved into cloud architecture at telecom scale. Since 2017 I have been working on cloud, data and AI platforms at a major entertainment company, leading architecture and engineering work at enterprise scale and building platforms that serve millions of guests. Along the way I picked up the certs people ask about (AWS Solutions Architect Professional, CKA, CKAD, RHCE, Microsoft DevOps Engineer Expert) and published two books.

The newest chapter is applied AI. I lead that work in my day job, and I build in public through Loki Mode and Autonomi. It is where everything above comes together: Linux operations, distributed systems, cloud platforms and a lot of hard lessons about what breaks in production.

Loki Mode / Autonomi is an autonomous software factory. You give it a GitHub issue, a spec or a one-line task, and it plans, builds and verifies, then opens a pull request with a receipt you can re-check offline. It runs on your machine with your own keys, and works with Claude, Codex and OpenCode. The source is available under BUSL-1.1.

The part I care about most is verification. I do not want an agent that says it is done, I want one that proves it.

npm install -g loki-mode
loki owner/repo#123     # turn a GitHub issue into a PR

I am also doing independent research on dependable multi-agent systems (HARLM, a paper on adaptive recursive language models) and finishing MIT Professional Education work in AI and machine learning. An MBA is next.

More on the portfolio page.

I have published two books, both in 2022:

• Mastering Cloud Engineering, a practical guide to building and running cloud infrastructure across AWS, Azure, GCP, Kubernetes and Terraform. (Amazon)
• From Strangers to Founders, stories of immigrants who helped build American industry and technology.

I also keep a blog with close to 200 posts, from RHCE and Docker 1.0 to re:Invent recaps, agents and what is different about leading AI teams. There is a newsletter too, Loki's Log.

Python, Go, TypeScript and Shell. AWS, GCP, Azure, Kubernetes, Docker, Terraform, ArgoCD and Jenkins. Lately: LLM orchestration, multi-agent systems and evaluation.

• Website and blog: asklokesh.com
• Autonomi: autonomi.dev
• Loki Mode: github.com/asklokesh/loki-mode
• Book a 30 minute call: asklokesh.com/schedule