I’m Aditya, a developer connecting thoughtful web experiences with the possibilities of AI.
From full-stack applications to explainable machine learning, my work spans the interface, the API and the systems behind them. I’m studying Computer Science with AI & Data Science at MIT World Peace University.
Away from the screen? A hike, a game of badminton, or a fresh page to draw on.
A selection of independent projects, research and work out in the world.
01 / MACHINE LEARNING2025–26
From raw data to models you can explain.
Behind the build
An end-to-end ML workflow platform covering preprocessing, feature engineering and selection, hyperparameter tuning, model comparison, experiment tracking and prediction. Redis queues and dedicated workers handle asynchronous training and explainability with stage-aware progress and ETA tracking. PostgreSQL stores metadata; bcrypt secures authentication. A Groq-powered assistant answers project-aware questions.
PYTHON / FASTAPI / STREAMLIT / REDIS / POSTGRESQL
02 / FULL-STACK APPLICATION2025
A clearer picture of everyday spending.
Behind the build
Expense management with secure Flask-Login authentication, CSRF protection, OCR receipt scanning, interactive dashboards and category analytics. Shared family workspaces combine role-based access, email invitations and CSV export with AI insights from a voice-enabled Gemini chatbot.
PYTHON / FLASK / POSTGRESQL / OCR / GEMINI
03 / RESEARCH & ACCESSIBILITY2024
Helping gestures become conversation.
Behind the build
Real-time fingerspelling recognition using a custom ResNet50V2 model, with 99.06% accuracy reported in the résumé (not an independent benchmark). LSTM sentiment analysis, translation into more than 30 languages, speech synthesis and transcript generation extend the recognition pipeline.
PYTHON / RESNET50V2 / LSTM / OPENCV
04 / CLIENT WORKPAID PROJECT
A digital home for people who make.
Behind the build
A paid website project for a Pune-based uniform and custom apparel manufacturer. The public website presents corporate, industrial, healthcare, hospitality and education apparel, with a catalogue, Apparel Studio and enquiry flow. The live site was inspected for these features; no conversion or performance improvements are claimed.