Blog

A comprehensive server combining voice control with motion recognition for sports analysis and fitness tracking

I built a typing practice app using AWS serverless architecture with a unique dual approach - microservices for production and monolith for development, all with CDK and TypeScript.

A comprehensive guide to streaming motion sensor data from microcontrollers to servers using smartphones as gateways, featuring React Native, Kotlin foreground services, and real-time data processing.

I scraped 1,700+ laptops using Python & BeautifulSoup, built statistical models achieving 93% R², and discovered graphics performance & processor specs drive pricing most.

I built a custom MTurk-survey integration using iframe + postMessage API for real-time data sync, with worker validation

In some cases, using voice or text commands to explore an app's capabilities is more intuitive than learning traditional navigation methods. I implemented this using multiple neural networks.

A shake-activated Bluetooth motion sensor device with adaptive sampling, efficient data streaming, and configurable parameters for motion tracking applications.

I built a 3D Renderer Web Application which dynamically and interactively visualises data without blocking the main thread.

Synthetic Data can be incredibly useful for training Neural Networks. Instead of relying on external Servers that you might have to pay for, you can use your own machine instead.

Creating Saliency Maps for Multi Head CNNs to highlight areas that were responsible for predictions

Translating natural language to SQL is not a task that requires external servers. A reasonably good consumer machine can handle the task just fine.

I used an Arduino Nano BLE Sense to detect keywords. Whenever the keyword was detected, the audio of the last second as well as the sensor data were stored

I created a custom PCB that lets me connect a microSD to a Seeed Studio Xiao nrf52

My goal was to train a neural network for Human Activity Recognition. I wanted to find an easy way of storing the data on a server without having to manually transfer the data while also being able to easily annotate the data.

My master thesis covered novel approaches for explaining Machine-Learning-Models

Existing Typing Trainers did not give me all the information about my typing performance I found interesting, so I created another app, which gives me more insights.

I wanted to create a website that intuitively shows how to perform tricks on a controller by using an animated SVG and GIFs