SYSTEMS BEYONDTHE DEMO.
I build the part people only notice when it fails: AI pipelines, secure APIs, automation, and backend systems that stay useful after the demo lights turn off.
I am building Vivid Store AI to turn ordinary CCTV into operational intelligence: movement tracking, unique visitor counts, camera-layout checks, and multi-camera reports teams can actually use.
The real work is not just detection. I am shaping the reliability layer around session isolation, duplicate suppression, tracker tuning, review states, and source-readiness checks so the output remains credible outside a controlled demo.
I like the uncomfortable layer where identity, latency, noisy inputs, persistence, and operational tradeoffs all show up at once. That is where software starts earning trust.
AI / Computer VisionI am building a retail intelligence platform that converts existing camera feeds into movement data, visitor counts, calibration workflows, and multi-camera operational reports.
My focus is the reliability layer: isolated sessions, camera mapping, duplicate suppression, human review, confidence flags, and workflows for both uploaded footage and live sources.
Secure backend + full-stack authI built an authentication platform around stateless access, OAuth2 social login, profile boundaries, and durable refresh-token rotation.
The architecture spans Spring Security filters, OAuth2 handlers, REST contracts, JPA persistence, and the React interface that exposes the account lifecycle cleanly.
Realtime backend systemI built a food-ordering and communication platform around concurrent WebSocket connections and low-latency multi-user flows.
I owned the Java and Spring Boot backend, tightened WebSocket handling under concurrent use, and carried the system through repository and deployment workflows.
I move across the full path from model output to database state, especially at the boundaries where systems usually get fragile. The experience can stay clean because the engineering underneath is doing its job.
I care less about AI as a buzzword and more about making models behave like dependable product features with clear inputs, guardrails, and outcomes.
I think in full pipelines: detection, tracking, calibration, validation, and human review working together instead of living as disconnected demos.
I build Spring Boot and FastAPI services with deliberate contracts, secure boundaries, durable persistence, and structure that can grow without becoming opaque.
I treat JWT, OAuth2, Spring Security, refresh-token rotation, and access boundaries as architecture - not as a checkbox before launch.
I design WebSocket-backed interactions around concurrency, predictable state, and communication that still behaves when more users arrive.
I shape schemas and persistence choices around access patterns, reporting needs, and operational reality - not whichever database is fashionable.
I take on a limited number of website builds for people who want more than a template. Positioning, interface, motion, and production code stay connected from the first screen to the final deploy. My business site, XDCoderz, is where that work turns into a focused service and product catalog.
I turn positioning, visual direction, responsive interaction, and production code into one coherent web presence.
I connect the interface to the systems that make it useful: secure APIs, authentication, databases, dashboards, and operational workflows.
I design AI assistants, search, automation, and computer-vision interfaces around a real job to be done - not AI for decoration.
My GitHub is less trophy shelf, more working notebook. It shows the experiments, architecture choices, and product ideas behind the polished work.
Computer-vision experiments and backend foundations behind my Vivid Store AI direction.
View repoA dual-agent engineering assistant that separates research, reasoning, and structured output.
View repoMy complete authentication lifecycle across React, Spring Security, JWT, and OAuth2.
View repoRealtime communication and food-ordering flows backed by WebSocket-driven interactions.
View repoAn exploration of AI-assisted logistics workflows and operational interface design.
View repoI use Java when structure and long-lived services matter, Python when AI and iteration speed lead, and everything else only when it makes the system clearer, faster, or easier to operate.
I am still early in the journey, but I am already doing the kind of work I want more of: technically demanding systems with visible product impact.
I work on automation, backend workflows, and AI-assisted product systems while learning how engineering decisions behave under real delivery constraints.
I delivered across the full stack and turned classroom fundamentals into working software, clearer implementation habits, and stronger ownership.
University of Lucknow - Expected graduation: 2027
The work that holds my attention rarely fits inside a neat label. It might begin with a model, run through an API, collide with imperfect data, and finally need an interface that makes the complexity feel simple. That full path is where I'm most useful.
Legacy JavaScript Algorithms and Data Structures - freeCodeCamp
Google Cloud Solution Challenge - Vertex AI strategy development
If the idea involves AI, backend architecture, automation, or a website that needs more than a template, send the context. I'll tell you where I can create leverage.