AI Systems · Backend Platforms · DataOps
I build production-grade systems — multi-agent LLM pipelines, 3D web experiences, data infrastructure, and full-stack products. End to end.
I'm a 2024 ECE graduate from NIT Srinagar, based in Bengaluru. My work spans AI application engineering, backend systems, data infrastructure, and frontend — I'm most comfortable when I own the full stack.
At Sigmoid Analytics, I completed a six-month structured training program across the modern data and DevOps stack, then built a production multi-agent LLM system from scratch in 24 days. Before that, I shipped three full-stack personal projects spanning a real-time social graph, a Stripe-integrated e-commerce platform, and a 3D configurator with physics simulation.
I care about systems that are observable, honest about their limits, and maintainable by people who weren't there when they were built.
Select a project to see the architecture and technical breakdown.
Intent router dispatching to 3 specialist agents — product search, order tracking, support escalation. Deterministic safety gates keep auth and cancellation logic in Python, never in the LLM. 5-step query relaxation, tier-based ranker, full observability. Built in 24 days.
Full social graph with six Prisma models, composite indexes, and a bidirectional follow system. Real-time notifications (LIKE / COMMENT / FOLLOW) via Server Actions. Clerk auth, UploadThing media, Radix UI components, dark/light theme.
Customer storefront, admin dashboard, and Express backend as three separate apps with distinct JWT auth layers. Stripe + COD payment flows with webhook verification. Cloudinary image hosting, full order lifecycle management.
Real-time 3D skateboard configurator with deck, wheel, truck, and bolt customization — all synced to URL params for shareable builds. HDR environment lighting, Matter.js physics simulation, GSAP scroll animations, Prismic CMS with 5 custom content types.
Two pipelines from Sigmoid internship. Medallion ETL (Bronze→Silver→Gold) on Databricks with PySpark and Delta Lake for ACID-guaranteed transforms. Metadata-driven CDC ingestion on Azure Data Factory with watermark-based change tracking and parameterized ForEach loops — one template handles N source tables.
Across AI systems, backend engineering, data infrastructure, DevOps, and frontend.
7 months at Sigmoid Analytics — structured training, then a production capstone.
A six-month structured engineering rotation across the modern cloud, data, and AI stack — followed by a solo production capstone taken from blank repo to a deployed, observable system.
Actively looking for roles in AI engineering, backend systems, or data engineering. If you're building something interesting, I want to hear about it.