Hello, I'm Aditya

Instrumems Gas Flow Sensor (Work Project), Altium, Kicad, Cypress PSOC Creator, STM32 Cube IDE 2021-2023
This project was an eval kit developed for the Instrumems Nano wire. This PCB biases the sensor, reads the output using an ADC and converts the raw voltages to flow values. Noise achieved was around 50uV or 0.5% of full range flow. The initial phase of this project that I inherited had a Cypress PSOC CY5866. I additionally developed firmware for this design. Due to supply chain issues during 2021, I designed and built a new PCB to replace this design, with a STM32 processor and migrated the firmware. An improved performance was measured compared to the older generation.

Instrumems Bubble Detection (Work Project), Altium 2021-2023
We use a hot wire anemometer to detect bubbles in a tube. The response of a hot wire looks different when it surrounds air or water. I developed hardware temperature compensation methods and built a PCB using it, optimizing the design for cost and performance, while conforming to application specific requirements.

Instrumems Multichannel CTA (Work Project), Altium 2021-2023
This was a showcase of the extremely fast frequency response of the hot wire anemometer. The instrumems nano wire is the industries fastest flow sensor. This CTA box contains 3 channels of extremely low noise and high frequency response anemometer circuits, which drives a neno wire probe connected with a co-ax cable. A frequency response of between 300kHz and 1MHz is achievable depending on the MEMS die used. I designed the PCBs, including the PSU and analog design, to provide ultra low noise flow readings. This box in combination with our sensor makes for the fastest flow sensor in the industry.

Instrumems Gas Detection (Work Project), Altium, python, 2021-2023
This was an attempt to using a hot wire anemometer for gas detection. The response of a hot wire depends on the thermal conductivity of the surrounding medium. I developed novel methods of biasing the sensor and analyzing its data. A detection accuracy of 200 ppm was achieved for gasses like Hydrogen. I built the EVK, designed the test setup and analyzed the output data.

Keep Car Alarm (Work Project), Altium, MATLAB, 2019-2021
A novel car alarm that detects break-ins in your car and acts as an active deterrant to crimes. I built custom PCBs, developed prototypes, and developed algorithms for very sensitive sensors.

Pressure Ulcer Prevention Project (Georgia Tech GRA Project), Autocad Eagle, Python, 2018-2019
A smart blanket that acts as a Ulcer Detection System among infants and handicapped patients. I designed a custom PCB for a DAQ system that took in data from pressure and moisture sensor array via a DAQ, with singal conditioning, data processing and pushing data to a cloud.

Doppler Radar Motion Detector, Altium, 2020
A 2-hour mini project for long range presence detection. I made this as a easy way to control the lights in my house as soon as the radar detects motion in a 3-5m area. Designed to be plugged into a wall outlet USB Charger. Soldered and tested this PCB myself.

Dual PIR Motion Detector, Altium, 2020
A 2-hour mini project for short range presence detection with direction detection. I made this as a way to detect presense with my home automation setup (HASS.io). Soldered and tested this PCB myself.

Design of Circuit Logic Simulator with Fault Simulation, C++, 2018
This simulator will simulate any electrical circuit given and can find the output given the input logic values. The simulator can also generate input test vectors to test any given stuck-at fault in the circuit or test the entire circuit. The algorithm used is Path Oriented Decision Making (PODEM).

Implementation of Cache policies in a Superscalar Processor, C++, 2017
I wrote code to implement various cache policies on a simulated SESC proessor. The performance and accuracy of each applied policy was compared using benchmarks.

Color 3D Printer / CNC machine, C++, PHP, JavaScript, 2017
Funded by the Department of Electronics, DJSCOE, Mumbai University.
We built a 3D printer capable of printing objects in full color. The project also had a CNC spindle that could help in PCB prototyping. The printer is funded by and currently installed in the EDM lab of our college. The printer has a website through which users can submit their models and start prints. I was incharge of firmware development and web development of the website. I was the Team Lead.

Wireless Speaker System using IOT, C++ 2017
Funded by the Department of Electronics, DJSCOE, Mumbai University.
This project was built into our college system as a efficient way for professors and staff to make announcements to our department. A web app was made to record announcements and send them to the college speakers from any web connected device.

Rubik's Cube Solver, Assembly x86, C++, 2016
Won 1st place at University level and 3rd place at a national level project display.
This was a robot that could solve any scrambled Rubik's Cube. A camera would scan the unsolved cube, compute a solution and solve the cube using servos. Coded in OpenCV, Atmega (Assembly) and Arduino, GUI made in Visual C++

Voice Controlled Home Automataion using Android, Android, Arduino, 2015
Developed an Android application that would use voice commands to control devices in the house. Communication would occur wirelessly over bluetooth. Written using Android and Arduino microcontrollers.

Automatic Determination of Cardiac Volumes from MRI Scans, Python (TensorFlow), CUDA, 2018
A group project designed to estimate the cardiac volume of a heart from a MRI scan, and to compare the accuracy of various machine learning methods used to estimate the same. Project Website, Git.

Stock Market Trading Agent, Python (TensorFlow), 2018
Using Machine Learning to make stock market predictions. Implemented and compared the performance of Conv. Neural Networks, Random Forest, Decision trees. The strategy provided a return of 6 times, outperforming the stock. Project implemented for CS 7646-Machine Learning for Trading. Code and description

Artificial Intelligence based Virtual TA, Python, 2018
This is an AI bot capable of learning given material and answering questions based on it. Given the syllabus of a course, this bot can learn the given concepts, make common-sense assumptions and answer questions students may have based on the given information.
Made for the course CS 7630: Knowledge based Artificial Intelligence.

Implementation of a Convolutional Neural Network on a GPU kernel, CUDA C++, 2018
Writing a kernel on the NVIDIA Kepler GPU, to implement a Neural network that performs basic OCR. This project involved writing CUDA kernels that perform functions like Convolution and ReLu on large arrays of data in parallel. Distribution of memory and kernel design were researched and optimized to provide maximum performance. Code written in CUDA C++. Git