●the lab is open
Jayakrishna Ichapurapu
AI Product Builder
Building multi-agent systems, developer tools and automation.
I build systems.
- 01AI
- 02Products
- 03Automation
- 04Developer tools
- 05Intelligent workflows
Not features in isolation — the whole loop. Ideas become systems, systems become products, products become experiments, experiments become the next idea.
02 — the lab
The Lab
Ideas → Systems → Products
Status labels are literal: Built means shipped code you can run. Exploring means research in progress.
03 — on the bench
Experiments
Smaller builds and the threads I keep pulling on. Repos link to code; interests are exactly that.
- 04Agent harnessesAGENTS.md-driven setups that keep agents honest — no fake tests, no branch hoppinginterest
- 05QE automationTest automation as part of the delivery system, not after itinterest
- 06Accessibility automationAccessibility IDs as the contract between design, app and testsinterest
- 07Git / PR automationValidation that runs where the work happensinterest
- 08Figma → app workflowsDesign components that carry their own test hooksinterest
04 — systems
How I think about software
Measurement feeds the next idea. The loop is the point.
05 — end to end
How I approach itOne workflow, not eight tools.
Design, code, accessibility, tests, monitoring and Git usually live in separate tools owned by separate teams. I like treating them as one pipeline where each stage hands the next a contract — accessibility IDs are the thread that lets a Figma component become something a test can find and a monitor can watch.
- 01Figma
- 02Component
- 03Application
- 04Accessibility IDs
- 05Automation tests
- 06Monitoring
- 07Git / PR
- 08Validation
06 — the ecosystem
GitHub, as a living system
Repositories are nodes, themes are clusters. Hover or focus a node to inspect it; select it to open the repo.
○ snapshot
Agents
07 — about
Who is Jayakrishna?
I like taking problems that normally need several disconnected tools — and several disconnected teams — and asking: what if the whole workflow could understand itself?
For a long time the collaborators were people. I've spent 12+ years in digital delivery — getting 15 app teams, around 400 people, to ship as one system, and connecting 7-Eleven to more than ten ordering and delivery platforms across two countries.
Now more of the collaborators are AI agents, and the problems look familiar: stale context, overlapping work, decisions nobody can trace. Neo, Braxis and AURA are what happens when I bring that coordination instinct to agents.
I started in biotechnology, which is probably why I still think in experiments. I build in the open, ship small, and label things honestly — built, exploring, or planned.
- based in
- McKinney, TX
- building
- multi-agent systems
- day job
- engineering leadership
08 — where the instinct came from
Problem → System → Impact
2018 — now
7-Eleven
Sr. Software Engineering Manager — Consumer Apps, Last-Mile Delivery, QE
Problem
Customers had to be able to order wherever they already were: 7-Eleven's own apps and third-party marketplaces it didn't control. 7-Eleven also had to support its own last-mile delivery platform and those third-party marketplaces at scale.
System
Delivery across 7-Eleven's consumer apps: the 7-Eleven Rewards app, the Speedway app and 7NOW, plus Skipcart, 7-Eleven's last-mile delivery platform. Marketplace integrations onboarded 10+ ordering and delivery platforms (DoorDash, Uber Eats, Grubhub, Postmates, Instacart, Google Food Ordering) across two countries.
Impact
- Supported 7-Eleven's mega portfolio of in-store and outside-store experiences, enabling delivery of millions of orders across two countries.
- CIO Recognition Award (July 2020) for showing 7-Eleven's leadership principles “Act Like an Entrepreneur” and “Challenge the Status Quo”.
- CDO Recognition Award (Nov 2019) for leading the marketplace integrations with external partners and delivering ahead of schedule.
- CEO recognition at a company town hall (July 2020) for delivering and supporting the store POS upgrade and the marketplace projects.
- Coordinates releases across 15 app teams, about 400 people.
2017 — 2018
The Walt Disney Company
Studio Owner — IA Cloud Platform
Problem
Disney was building its first cloud-agnostic platform for taking in, storing and transforming data, and every application team needed a reliable way to move onto it.
System
Owned the onboarding pipeline. Prioritized deployment and support requests, ran weekly syncs with the teams being onboarded, and documented the onboarding process every team followed.
Impact
Onboarded the first project onto Disney's first cloud-agnostic data platform, then the projects that followed it.
2016 — 2017
Deep North
Delivery Lead — AI / Computer Vision
Problem
A computer-vision startup had working research (face recognition, people counting, tracking across cameras, fall detection) that had to become products customers could run on site.
System
Led delivery with engineering teams in the US, China and India: requirements, deployments at client sites, quality checks, and the web analytics dashboard. Worked with the CEO and VP of Sales on timelines and pre-sales.
Impact
Delivered the people-detection service end to end for major retail and entertainment customers, and ran the live audience analytics at Microsoft Ignite in Germany and the US.
2014 — 2016
Earlier
LessAccounting and Qount
Led delivery and releases for two small-business accounting software products, from gathering requirements through rollout.
Gilead Sciences
Tested and documented software used in pharmaceutical work.