Aditya Chaudhary

A production supplier-chain observability platform built for SLR Consulting / RCS Global, enabling analysis, visualization, and monitoring of critical-mineral supply chains and relationships between suppliers.

Contributed primarily to the backend modernization of the platform, migrating the application from a legacy backend to a modern Python/FastAPI stack while preserving existing business workflows and application functionality.

Led the migration of existing backend functionality by redeveloping legacy APIs with FastAPI, restructuring backend components, and implementing new APIs and supplier-chain features required by the modernized platform.

Worked extensively on database migration and data-model evolution, adapting legacy schemas and migrating existing application data to the new backend architecture while maintaining data integrity and compatibility with existing workflows.

Implemented and maintained data-processing pipelines and backend integrations for supplier-chain data, including validation, transformation, and application-level processing required to support the platform's observability workflows.

Developed comprehensive unit, integration, and API test coverage using Pytest and pytest-asyncio, providing regression protection throughout the migration and subsequent feature development.

Containerized backend services using Docker and maintained OpenAPI/Swagger API documentation, supporting development, integration, deployment, and ongoing maintainability of the platform.

Contributed to a large-scale multi-tenant learning platform under the XPRIZE ecosystem, supporting scalable and content-driven education workflows.

Worked on a microservices-based architecture with schema-based tenant isolation, enabling secure and efficient tenant-specific data handling.

Designed and developed core platform capabilities including learning workflows, progress tracking, content management, and RBAC-based access control.

Designed Retrieval-Augmented Generation (RAG) workflows using LangChain, embeddings, vector search, and LLM integration to generate context-aware responses grounded in course materials and learning documentation.

Developed document ingestion and indexing pipelines for learning resources, including text extraction, chunking, embedding generation, and semantic retrieval.

Integrated LLM tool calling with backend APIs to retrieve learner progress, course details, and learning resources, combining structured application data with retrieved knowledge.

Implemented tenant-aware retrieval and RBAC enforcement to ensure AI-generated responses respected course-level permissions and tenant data isolation.

Designed event-driven asynchronous workflows for background processing and AI document indexing, using background jobs and Redis to support scalable content updates and responsive services.

Optimized performance using Redis for tenant-aware configuration caching and efficient request handling.

Enhanced platform security with file validation pipelines and region-specific access control mechanisms.

Contributed to infrastructure using Docker, Kubernetes (AWS EKS), and Codefresh CI/CD pipelines across staging, demo, and production environments.

A production-scale video-based social engagement platform serving 10,000+ users, focused on content discovery and competition workflows across multiple levels.

Implemented CQRS architecture by separating read and write models, leveraging a denormalized read store with background reindexing on data mutations to eliminate complex joins and significantly improve query performance.

Designed an event-driven architecture to decouple core modules, enabling asynchronous workflows such as email notifications, CSV imports, cleanup tasks, and search reindexing using Celery and message queues.

Built a scalable notification system capable of handling 10,000+ email messages using Amazon SQS with a worker-based architecture supporting dynamic upscaling and downscaling.

Implemented retry mechanisms with retry queues and dead-letter queues (max 3 attempts), improving reliability and failure handling across distributed workflows.

A multi-tenant SaaS platform for local businesses that enables vendors to create professional digital storefronts with custom branding, products, collections, business information, vendor-specific subdomains, and direct WhatsApp ordering.

Built the core management platform using Django REST Framework and a React + Vite dashboard, implementing vendor onboarding, authentication, product and collection management, portfolio customization, publishing workflows, access control, and subscription management.

Designed a distributed read architecture where AWS Lambda generates versioned vendor-specific SQLite read databases from PostgreSQL, uploads them to Amazon S3, and publishes Redis events to notify the FastAPI portfolio service when new data is available.

Built a lightweight FastAPI read layer that consumes Redis events and performs atomic SQLite database replacement. Implemented version-based caching for fast vendor portfolio responses using nginx and used S3 + CloudFront for CDN-based delivery of the public React application.

Implemented live Razorpay subscription and payment processing with webhook handling and payment verification, along with Nginx as a reverse proxy and CloudWatch for application and infrastructure logging.

Deployed the platform using Docker and GitHub Actions CI/CD across AWS infrastructure including EC2, S3, CloudFront, Lambda, PostgreSQL, Redis, and CloudWatch.

This is a full-stack note-taking application built using Django (DRF) for the backend and ReactJS for the frontend. It features secure authentication using JWT with registration, login, logout, token management, automatic token refresh, and blacklist handling for enhanced security. Users can seamlessly perform full CRUD operations on notes including create, edit, delete, pin, and unpin. A key feature of the platform is the ability to generate public shareable links for notes. Users can mark a note as public, which generates a unique link that can be shared on social media or with anyone. Recipients can view the note through the public archive page without needing an account. Notes can also be switched back to private at any time for full control over visibility. Deployment is automated using GitHub Actions CI/CD. The backend runs on AWS EC2 using Gunicorn and Nginx, with DuckDNS for domain routing and Let's Encrypt for HTTPS security. The frontend is hosted on Netlify, and the database is managed on NeonDB for reliability and performance.

Developed a Django-based social media platform enabling users to create, like, comment, and save tweets, with a follow/unfollow system and real-time notifications for activities like new followers, unfollowing, and tweet interactions. Built a secure authentication system with registration, login, logout, password reset, and profile management for updating profiles, uploading pictures, and managing tweets.
Added media uploads for tweets and profile images to ensure a rich user experience. Enhanced interactivity with a modern UI, Bootstrap styling, and dynamic Toast Messages for real-time feedback (e.g., login success, new comments, likes, or errors).

Developed a file sharing platform allowing users to upload, manage and download resources securely. Implemented authentication, integrated AWS S3 cloud storage for file management and used PostgreSQL (NeonDB) for database management. Application deployed using Vercel.

A collection of mini projects built to demonstrate core JavaScript and DOM manipulation.

Includes projects such as weather finder, dynamic user cards, news web app, to-do list with local storage, image carousel, number guessing game and more.