I’m a pre-final year B.Tech IT student at VIT Vellore with a strong interest in cloud computing, ML, cross-platform app development, and backend systems. Over the last couple of years, I’ve worked on projects ranging from mobile apps in Flutter to cloud-native pipelines on AWS, while also gaining hands-on experience with backend and ML frameworks.

Languages & Web: C++, Python, Dart, SQL, HTML, CSS

Cloud: AWS (SageMaker AI, EC2, S3, Lambda, Glue, Athena, IAM)

Backend & Databases: Express.js, PostgreSQL, SQLPlus

Frameworks & Libraries: Flutter, Pandas, Matplotlib

Tools: Power BI, Postman, Git, Android Studio

Certifications: AWS Certified Cloud Practitioner - Credential, Generative AI using IBM Watsonx - Credential

Production-style ML (including ETL) pipeline that ingests flight-delay CSVs and implements a medallion architecture (Bronze - Silver - Gold) on S3 to enable reliable analytics and inference.

Highlights:

• Medallion architecture: Bronze - Silver - Gold layers on S3 for incremental refinement, data quality, and easier downstream consumption.
• Cost-efficient querying: Athena CTAS/UNLOAD workflows using Parquet formats and partitioning for fast, low-cost queries.
• Modeling & serving: XGBoost classifier (artifacts stored in S3) and deployed real-time inference via FastAPI, containerized with Docker, with an Nginx frontend on EC2. (Serverless variant repo).
• Analyst access: Exposed Gold tables to Power BI via Athena using the ODBC driver for self-service BI and reporting.

Tech Used: AWS S3, Athena, EC2, XGBoost, FastAPI, Docker, Nginx, Power BI (ODBC)

Official backend for Hackulus’25, SIAM-VIT’s flagship hackathon with 300+ participants.

Highlights:

• REST API Development: Designed and deployed secure endpoints for managing tracks, submissions, and admin workflows.
• High Reliability: Achieved >99% uptime and minimal runtime errors during the event through robust system design.
• Authentication & Validation: Implemented JWT-based authentication, comprehensive input validation, and structured error handling.
• Deployment & Scalability: Deployed on Render for real-time performance and smooth scalability during the hackathon.

Tech Used: Node.js + Express.js (backend), PostgreSQL, REST APIs, JWT Authentication, Render

A smart photo vault app built with Flutter that automatically organizes images by location metadata.

Key Features:

• Auto-organization: Photos sorted into folders by city.
• Search: Retrieve images instantly with place-based keywords.
• Interactive Map: Explore all saved photos on a Google Maps interface.
• Themes: Light/dark theme support with persistent user preferences.

Tech Used: Flutter, image_picker, geolocator, Google Maps API, path_provider

ML system that predicts the most suitable catalyst for a given chemical reaction using graph-based molecular representations.

Highlights:

• Graph-Based Machine Learning: Utilizes Morgan fingerprints (via RDKit) to encode reaction SMILES from the ORDerly dataset.
• Full Preprocessing Pipeline: Includes reaction cleaning, normalization, agent explosion, data balancing & malformed-line handling.
• Model Training Suite: Custom training pipeline with adjustable fingerprint size, class weights, chunked loading and joblib-based artifacts.
• API + Frontend: FastAPI backend with a minimalist frontend for quick catalyst predictions.
• Local Debugging: CLI inference script for rapid testing with real Reaction SMILES.

Tech Used: RDKit, Scikit-learn, FastAPI, Uvicorn, Pandas, NumPy, Joblib, Morgan Fingerprints, ORDerly Dataset

A cross-platform app that lets users transfer playlists between Spotify, YouTube Music, and SoundCloud. (In progress)

Highlights:

• Multi-platform Support: Secure auth and playlist selection for all three music services.
• Filtering & Transfer: Easy-to-use UI for selecting and transferring playlists.
• Backend: REST API built with Express.js + PostgreSQL for secure handling.
• Status Tracking: Displays auth and sync status for each music service.

Tech Used: Flutter, Node.js + Express.js (backend), PostgreSQL, REST APIs

Kindly visit my repositories for more such projects.

• Email: dshryng@gmail.com
• LinkedIn: Aryan Deshpande