Architecture trade-off
Taking Carrier APIs Off the Prediction Critical Path
Circuit breakers kept checkout fast, but they were quietly dropping the best carriers from the mix. I replaced real-time carrier calls with daily precomputed data.
Writing
Architecture Notes
The real stories behind the numbers: what the problem was, what I did about it, and what it was worth. Short version up top, full story one click away.
Start here
About Srinivas
14+ years, one obsession: build the same delivery confidence Amazon Prime gave shoppers, for every e-commerce brand.
Architecture across every discipline
Architecture Across UI, Data, and Backend Boundaries
I lead the UI/UX, data/ML, and backend teams as one architecture, and with AI-assisted tools, I build across all three myself.
Ownership & closing power
Using Architecture to Close a Four-Year Enterprise Opportunity
A 4-year-pursued enterprise account finally closed, because of a 2-month build, not a bigger pitch deck.
Product thinking, not just engineering
Turning Post-Purchase Pages Into a Revenue and Retention Channel
A tracking page, a notification system, and a coupon engine: a 5-9% conversion lift and a 95% drop in support calls, from pages that used to have neither purpose.
Agentic AI, shipped
Turning Institutional Knowledge Into a Tool, Not a Wiki Page
Four RAG and agentic copilots, all live in production for internal teams, turning documentation, past resolutions, and static pages into something you can just ask.
Independent problem-solving
Closing Shopify's Order-to-Prediction Attribution Gap
Shopify gives you no way to connect an order back to the prediction that led to it. I built one anyway, at 92% accuracy.
Root-cause diagnosis
Eliminating Silent Data Loss in Event-Driven Order Processing
A failure mode invisible on both ends at once: diagnosed, rebuilt event-driven, and never recurred.
Requirements from the source
Designing 3PL Onboarding by Starting at the Warehouse, Not the Whiteboard
I visited a major 3PL's warehouse to understand their problem firsthand, then designed and built an entirely separate product to onboard them and their 100+ brands.
Durable architecture
Designing a Canonical Data Model That Survived Platform Expansion
One data structure normalizing every source and every carrier, unchanged since I built it, and still what every product runs on.
Reliability engineering
Building Observability That Catches Failures Before Customers Do
Full-stack observability plus a daily data-integrity monitor: most issues get found and fixed before a single customer complaint comes in.
Platform-wide optimization
Removing the Thread-Cost Ceiling During a Java 21 Modernization
A Java 21 migration to virtual threads, removing a thread-cost ceiling that touched order processing, tracking, and every tenant's notifications at once.
Cross-functional impact
Making the Business Case With a Prospect's Own Data, Not Slides
I built a way to walk every prospect through their own numbers instead of a generic deck, and conversion improved significantly.
Beyond the architecture diagram
Owning the Customer Relationship, Start to Finish
I lead onboarding and Level 1 support too: 60+ retailers onboarded, plus a 100+ store enterprise 3PL, and that trust is what turns a single-product account into a multi-product one.
Zero to production
Zero to Production: Building a Platform From an Empty Repo
From an empty repo to our first customer to a Shopify App Store listing: I built the foundation everything else was built on.