FULL-STACK AI ENGINEER · REACT → FASTAPI → RAG

I build intelligent products from interface to inference.

I’m Vishnu Murti Pandey, a FullStack Engineer who builds complete AI products: refined React experiences, Python APIs, retrieval systems, streaming workflows, and deployable infrastructure.

request.pipeline
Systems operationallatency-aware
01React UI
02FastAPI
03RAG
04LLM
05SSE
async def answer(query):
 context = await retriever.search(query)
 yield agent.stream(query, context)

01 / ABOUT

Frontend roots. Full-stack range. AI focus.

I started with responsive, accessible interfaces in React and Next.js. As products became more data- and intelligence-driven, I moved deeper into APIs, databases, retrieval, model orchestration, evaluation, and deployment, without leaving product quality behind.

InterfaceThe product people use
SystemsThe logic that makes it reliable
IntelligenceThe context that makes it useful

02 / EXPERIENCE

Building across the application stack.

My AI work leads the story; my Magic EdTech experience remains the full-stack and frontend foundation beneath it.

01

Sep 2025 — Present

Applied AI · Full stack

Senior Software Engineer

Hughes Systique Corporation

Building DocuMind, a production-oriented multimodal RAG platform that turns research documents into contextual, cited conversations.

  • Engineered an asynchronous pipeline spanning PDF parsing, semantic chunking, image and table enrichment, hybrid retrieval, reranking, and prompt construction.
  • Built SSE-streamed responses with citations, figures, and tables using FastAPI, PostgreSQL, Qdrant, MinIO, and OpenAI.
  • Architected reusable React and Next.js modules with TypeScript, Context API, Tailwind CSS, REST integrations, and performance-conscious rendering.
02

Jun 2022 — Aug 2025

Full stack · Frontend foundation

Software Engineer · Full-Stack Developer

Magic EdTech

Built product workflows end to end, with strong ownership of responsive interfaces, authentication, data-heavy dashboards, and backend integration.

  • Led development of a deal-pricing platform serving client and vendor workflows, improving operational visibility across the product.
  • Delivered dashboards, profile and form experiences, data grids, charts, and responsive interfaces using modern React engineering practices.
  • Designed and integrated Google authentication while contributing across Node.js, Express, APIs, and application data flows.

03 / CAPABILITIES

One engineer, connected layers.

Organized by how I ship products not by a wall of logos.

01

Production experience

Interface

ReactNext.jsTypeScriptReduxTailwind CSSAccessibilityPerformanceStreaming UI
02

Built with

Intelligence

LLM applicationsMultimodal RAGEmbeddingsHybrid retrievalRerankingRAGASDeepEvalPrompt design
03

Built with

Systems

PythonFastAPINode.jsExpressPostgreSQLMongoDBQdrantREST APIsSSEJWT
04

Working knowledge

Delivery

DockerMinIOGitVercelRenderTestingSystem designCloud-ready architecture

04 / SELECTED SYSTEMS

Products with real engineering depth.

The strongest work first: architecture, product thinking, and implementation decisions in one view.

01

Multimodal RAG platform

DocuMind

Professional work

Document-aware AI conversations grounded in uploaded research, with citations, figures, tables, and persisted memory.

ENGINEERING CHALLENGE

Coordinating asynchronous ingestion, enriched context, retrieval quality, and a responsive streamed interface as one coherent system.

ARCHITECTURENext.js → FastAPI → PostgreSQL / MinIO → Qdrant → reranking → OpenAI → SSE
02

Real-time full-stack application

Chit-Chat

Live project

One-to-one and group messaging with authenticated users, participant management, and real-time communication.

ENGINEERING CHALLENGE

Keeping client state and group membership synchronized across authenticated real-time sessions.

ARCHITECTUREReact → Express / JWT → MongoDB → Socket.io
03

Realtime canvas system

Collaborative Sketch Board

Live project

A shared digital whiteboard with drawing tools, undo/redo, sticky notes, canvas upload, and download.

ENGINEERING CHALLENGE

Synchronizing user actions over sockets while maintaining an intuitive, low-friction canvas experience.

ARCHITECTURECanvas UI → Node.js → Socket.io → connected collaborators
04

Browser spreadsheet application

Excel Clone

Live project

A browser-based spreadsheet interface for working with editable cells and familiar spreadsheet interactions.

ENGINEERING CHALLENGE

Translating desktop spreadsheet behavior into a responsive browser interface using core web technologies.

ARCHITECTUREInteractive grid → JavaScript state → browser rendering
05

Browser media application

WebCam App

Live project

A focused web application that works with webcam media through a lightweight browser interface.

ENGINEERING CHALLENGE

Handling browser media access while keeping camera interactions clear and straightforward for users.

ARCHITECTUREBrowser camera API → JavaScript controls → media interface

05 / CONTACT

Let’s build something intelligent.

I’m interested in opportunities where frontend engineering, backend systems, and applied AI come together to solve real product problems.