Hi, I'm Jigna Modi

My Story

Ten years at Amazon. Building what customers touch and what engineers build with, and the interesting problems have usually been where those two meet. The work spans seller listing and inventory systems, product catalog quality, enterprise employee data platforms, developer tooling, accessibility engineering, and AI-native front-end authoring — and what connects all of it is a habit of zooming out far enough to pick the right problem and then zooming in far enough to ship the solution.

The initial projects were about learning the ropes in data pipelines, catalog quality, and operational rigor. From there, the documentation platform is where things got interesting — a platform serving 40+ teams that was suffering from its own success. Slow manual onboarding, high publishing latency, no search integration, growing operational cost — and the right answer turned out to be not building something new but deprecating it entirely. After evaluating five alternative platforms and prototyping integrations to validate assumptions, the call was to migrate onto an existing hosting platform that already had 10+ aligned capabilities. A deliberate “reuse over reinvent” bet: 20 extra engineering days to consolidate two redundant products into one and eliminate ongoing dual-maintenance. With a team of four engineers the work delivered 600% adoption growth, brought onboarding from days to under 30 minutes, and migrated 40+ documentation sites via one-click automation with zero downtime. A cross-build-system deployment tool built along the way became reusable by other teams, and the ownership model negotiated with the host platform team is still the operating agreement today.

The pattern was becoming clear: capability that lives inside the workflow gets adopted; capability that sits outside it doesn't. Accessibility was the next place to test that. The interesting part about this chapter was the shift it enabled: an entire ecommerce organization of 100+ teams moved from reactive manual accessibility fixes to proactive AI-powered prevention. The architecture embedded accessibility agents throughout the software development lifecycle — from code authoring in IDEs to automated testing in CI/CD pipelines — catching violations at the point of creation instead of after they reached production. Shifting left like this meant fewer costly production remediations and a fundamentally different relationship between engineering teams and accessibility compliance.

The MCP and agentic patterns from that accessibility work carried directly into the current chapter: leading the vision, architecture, and hands-on delivery of an AI-native development platform for a storefront rendering ecosystem. The problem: builders want to go from intent to production-ready customer experience without leaving their AI assistant — describe what they want, get shippable code. But they build across multiple proprietary rendering frameworks that don't exist in any model's training data, every team solves the same onboarding and code generation problems independently, and nothing validates that generated code actually meets compliance and accessibility standards before it ships. An MCP-backed knowledge platform with per-framework grounding collapsed feature delivery from 3–4 weeks to 2–3 days for 200+ engineers. On top of that sits an open agent registry where teams publish, fork, and share composable agentic workflows — so one team's codegen solution is discoverable and reusable by the next instead of rebuilt from scratch. What started for one framework now spans the multi-framework rendering stack, positioning the platform as the default AI authoring layer for the entire storefront engineering organization.

The work that matters most to me is the kind where you ship something and then watch other people build things on top of it that you never imagined. That's the signal that it was the right abstraction.