Software Engineer & Solutions Engineer with 12+ years of experience building, deploying, and integrating production-ready systems across web platforms, APIs, and distributed architectures.
I specialize in end-to-end delivery — from system design and backend architecture to frontend UX, cloud deployment, AI, and third-party integrations.
📍 Toronto, Canada 🇨🇦
🌎 Open to opportunities in Canada & the United States
Next.js · Express · PostgreSQL · PayPal · Docker · AWS
- Production-grade publishing platform with admin workflows
- Secure one-time payments via PayPal
- Deployed on AWS EC2 with Docker and Nginx reverse proxy
- Real-world client delivery (architecture & documentation only)
➡️ Repository: Publishing-Platform-Architecture
Node.js · Express · RabbitMQ · Redis · Docker
- API Gateway with clear service boundaries
- Event-driven communication using RabbitMQ
- Redis-based caching layer
- Designed for scalability, integrations, and onboarding
➡️ Repository: MicroServices-Backend
Django · PostgreSQL · PayPal
- Full product catalog, cart, checkout, and order lifecycle
- PayPal payment integration (sandbox + live-ready flow)
- Modular, reusable architecture for real deployments
➡️ Repository: E-Commerce-Platform-Template
C++ · POSIX Sockets · Multithreading
- Concurrent multi-client file server
- Binary file transfer with chunking
- Focus on low-level networking and performance
➡️ Repository: FileServer
Python · LangGraph · LangChain · Groq (Llama 3.1) · Tavily Search · Pydantic
- Designed a multi-agent orchestration system with a Supervisor node that dynamically routes tasks across specialized agents (Prompt Enhancer, Researcher, Coder)
- Built a Validator agent as a quality gate that evaluates responses and loops back only on fundamentally incorrect outputs — minimizing redundant cycles
- Integrated Tavily web search and Python REPL tool as agent-bound capabilities for grounded research and live code execution
- Leveraged Pydantic structured outputs with
with_structured_output()for type-safe, deterministic routing decisions across agent transitions
➡️ Repository: LangGraph-Multi-Agent-Workflow
Python · LangChain · LangGraph · Vector Databases · Embeddings
- Designed and implemented a multi-step Retrieval-Augmented Generation pipeline with query decomposition, iterative retrieval, and cross-step synthesis
- Built vector-based retrieval over embedded document chunks with re-ranking and context aggregation to improve answer quality for complex, multi-hop queries
- Implemented structured LLM outputs for deterministic parsing and reliable downstream processing
➡️ Repository: RAG-LangGraph-Multi-Step-Pipeline
Book-Management-REST-API— Backend fundamentals and API design patternsAI-Recipe-Generator-Claude-API— LLM-powered web applicationAI-Content-Generator— AI-driven content generation platformVideo-Streaming-Platform— Netflix inspired movie-streaming web applicationShopping-Platform-Template— Amazon inspired frontend e-commerce platformDjango-Auction-Platform— Auction platform for selling and buying artifactsDjango-Wikipedia-Platform— Encyclopedia platform for uploading and editing documents
Languages
TypeScript, JavaScript, Python, C++
Backend & APIs
Node.js, Express, Django, FastAPI
REST APIs, Webhooks, OpenAPI / Swagger
AI & LLM Systems
LangChain, LangGraph, MCP
Embeddings, Vector Databases
RAG Pipelines
Databases & Caching
PostgreSQL, MongoDB, Redis
DevOps & Infrastructure
Docker, Docker Compose, AWS (EC2), Nginx
Tools & Integrations
Postman, PayPal APIs, CI/CD workflows
- LinkedIn: https://www.linkedin.com/in/upnitbanga
- Portfolio: https://upnit-portfolio.vercel.app
