Aman Tahiliani · Atlanta
Software engineer. I build agent infrastructure, pipelines that carry an idea from issue to reviewed, merged pull request, with humans at the gates that matter. Healthcare is where I learned to do it under real constraints: data that can't leave the VPC, a small shared AI budget, and 61 production instances that can't break.
I got tired of babysitting them one prompt at a time, so I built a pipeline that treats them like contractors: a spec they can't misread, an isolated workspace, a build gate before anything earns a pull request, and a reviewer that is never the agent that wrote the code.
Six harnesses plug in through one adapter function each, and that is the only tool-specific code in the system. One is currently disabled, because when I tested its headless path it wasn't reliable enough to trust with unattended work. That row stays in the table, with the date on it.
How the factory worksAgents that read the ticket and the architectural decision record before they touch code. Isolated multi-repo worktrees so three tickets run at once without their migrations colliding. Scheduled jobs that write my standup notes and my sprint demo docs. A Playwright gate that lowered the share of my tickets QA sends back.
An F1 race-weekend command center in Go and React. Live timing bridged from the official SignalR feed to the browser, and a local SQLite history that works when nothing is live, because for most of the year, nothing is.
A Raspberry Pi and a mini PC that don't just run services, they run agents. Scheduled agentic workflows I can drive from my phone, and a resident agent on Discord that has been awake longer than any chat session.
Open source
Updated
I extended this one rather than starting it: an Orca surface alongside the existing mux panes, and routing guidance so a subagent gets a model that suits the job.
The public evidence trail for my issue-to-merged-PR agent pipeline.
I contributed upstream to the OpenCode interface I run on my homelab.
A small Go service that keeps my homelab DNS pointed at home.
My public SMART on FHIR sandbox for the healthcare integration work I know firsthand.
I continued with the platform after HRS acquired Rimidi on 1 March 2026, supporting its ongoing development, helping teammates build context, and contributing to the architecture for bringing the two products together. I also run each sprint release across 61 AWS ECS production instances through Jenkins.
Before that: FHIR pipelines standardizing Epic and Cerner data for 20,000+ patients a day, a Georgia Tech MSCS, and four published papers.
Full history