AI-assisted engineering enablement
I set up Claude Code, Copilot and Codex so they make a team faster without making it sloppier: project instructions, custom Skills, MCP integrations, review subagents, and a usage policy people actually follow.
I'm an Associate Team Lead who designs .NET and Azure systems for healthcare and commerce, and helps engineering teams ship with AI without lowering the bar on correctness.
In 8+ years I've synchronised 100,000+ patient records across HIPAA-compliant integrations, cut server costs by 30%, and architected an EHR platform that reached 92% provider adoption.
Today I'm leading my team's adoption of Claude Code, GitHub Copilot and Codex, backed by a written policy, a 15-module curriculum and a reference application that proves the workflow.
I started in 2015 in Srinagar, building ASP.NET MVC sites and writing a foreach loop for everything. Then I found LINQ, and it changed how I think: describe what you want, and let the code read like the intent.
Most of my career since has been in healthcare. I architected Invaryant, an HL7-based Electronic Health Record platform for a US client, and later connected three separate healthcare portals to it while keeping every flow HIPAA-compliant. That work taught me to care about the boring parts: role-based access, data contracts, audit trails, and what happens on the 100,000th record.
In 2026 I moved into leading a team, right as AI started writing a large share of our code. AI-generated code tends to look clean, and clean code is no longer proof of correct code. So I wrote our AI-assisted development policy, built a curriculum around verification, and shipped a full application with Claude Code to show the team where review agents earn their keep. They caught six real bugs that a human reviewer would have waved through.
Outside the codebase, I run a commercial ultra-high-density apple orchard in Pampore. Farming and software reward the same habit: measure first, then change one thing at a time.
I set up Claude Code, Copilot and Codex so they make a team faster without making it sloppier: project instructions, custom Skills, MCP integrations, review subagents, and a usage policy people actually follow.
APIs that stay readable as they grow. I use CQRS with MediatR and Vertical Slice Architecture so each feature lives in one place, and I tune the data layer for cost and latency.
HL7 data flows, role-based authentication and HIPAA-compliant synchronisation between EHRs and patient portals, built for data you can't afford to get wrong.
From commit to production on Azure with one pipeline: App Service for APIs, Static Web Apps for the front end, Azure SQL, and GitHub Actions gluing it together.
Full-stack project management platform built end-to-end with Claude Code
My team was starting to use AI coding tools daily, with no shared rules and no evidence of whether AI-written code could hold up in production.
Build a realistic reference application entirely with Claude Code, and use it to show the team a safe, repeatable workflow.
HL7-based Electronic Health Record platform, later connected to three patient portals
A US healthcare client needed an Electronic Health Record system, and later had three separate portals holding patient data that had to stay consistent with it.
Architect the EHR, then integrate all three portals with it under HIPAA, without losing or corrupting records.
Moving a growing platform to a scalable, API-centric infrastructure
As usage grew, the platform's hosting bill and response times were climbing together.
Rework the backend so it scaled with load while costing less to run and responding faster.
Brought an ASP.NET Core Web API into a legacy store, reached SOC2 compliance, and lifted conversion by 15% through a reworked checkout.
Live-video product inspections that replaced site visits, saving 25+ hours every week.
Self-assessment and third-party review platform that cut audit time by 30% across 35+ factories.
Event-driven e-commerce demo with an in-memory event bus for orders, inventory, payments and notifications.
AiNewsSummarizerC# · .NET 9Web API that scrapes a news article and returns a three-point summary through the OpenAI API.
ImageResizerC# · .NET 8Resizes and compresses uploads to fit strict limits (≤100 KB) for government portals, using ImageSharp.
Leading a .NET/Azure team; authored the AI-Assisted Development Policy; running the 15-module AI engineering curriculum.
Integrated three healthcare portals with Invaryant under HIPAA; architected role-based auth and HL7 data flows for a US EHR platform.
Architected Invaryant; re-built the API layer for 30% lower server cost and 50% faster responses.
Client web applications in ASP.NET MVC and React.
Web applications on the ASP.NET MVC framework.
I write on LinkedIn for 2,000+ followers about .NET performance and engineering with AI.
Load reference data once, freeze it, and get fast, thread-safe, read-only lookups.
Why reviewers should ask "how do we know this is right?" once AI writes the diff.
Stop copying every JSON payload just to hand an API a Stream.
I'm open to Team Lead, Solutions Architect and Software Architect roles, and to conversations about rolling out AI-assisted engineering safely. Email is the fastest way to reach me.