Syed Rayyan Ahmed

Computer Graphics Engineer

• Using Computer Graphics Engine, we developed and released a toolkit that generates data for ML training this toolkit intends to generate quick customizable Synthetic Data.
• Enabled the proper working of perception package (toolkit for sim2real training) with HDRP.
• Developed Domain Randomizations for generating randomized data for different ML use cases.
• Investigated and developed a demo project for Synthetic data generation using Nvidia Omniverse (USD based graphics engine).
• Applied proper Labelling such as 2D Bounding box, Human keypoints, Semantic segmentations and Instance segmentation on generated data.
• Installed a Visualizer tool that helps visualize the labelling on generated datasets.

Game Developer & Artificial Intelligence Developer

• Worked with companies like Mahindra & Saudi Aramco for Augmented Reality and Virtual Reality projects, where Mahindra project was presented in Delhi Auto expo 2020.
• Worked on a Humanoid assistant robot with key software features like, natural language processing, robot localization, face scanning & identification and parallel processing.
• Developed a “temperature sensing face recognition based attendance marking system”.

Client Relationship Manager

• Designed plans to address client’s business needs.
• Researched the market and implemented new ways to retain client’s.

Web App - Black and White to Colour - Using Django

• As shown in the previous project (toColour), we know that a black and white image and video can be converted to a colored version.
• Machine learning projects like these have many real-world applications and can be very helpful for society, but their implementation is heavily dependent on hardware resources because machine learning requires a lot of GPU resources.
• What if we can upload these machine learning implementations on a cloud resource and provide easy access to it via a web app ?.
• Using the Django web framework, I implemented a Python-based web app, which allows users to access the black and white to color implementation remotely.
• Just like React.js or Node.js, Django is a web framework that follows the MVT (model-view-template) architectural pattern with key features such as high security, versatility, and scalability.

toColour - Black and White to Colour photo/video convertor

• toColour is an ML-based Python project.
• A DNN model is trained with various datasets, and the output of that training is saved as a Caffemodel.
• Using this model, we can convert an RGB image to a colored image.
• If we extend this idea even further, we can also convert a black and white video to a colored version.
• The video is a combination of many frames. From the input video, each frame is extracted sequentially, and that frame is colored and saved in an output location. Later, these colored frames are converted back into a video.

Customer Service Robot

• For Delhi Auto expo 2020 we were asked to develop this customer service Humanoid robot by Mahindra group with key software features like, natural language processing, robot localization, face scanning & identification and parallel processing.
• My main responsibilities was to develop all the software and operating system related functionality which include Computer Vision, Sensors, Voice Assistant, Face Recognition, Emotion recognition and Multiple Processing.
• Being a person from software background with a very little knowledge about electronics it has been a challenging experience for me. It took us months to develop this Robot and with every new steps and challenges we learned something new.

Make your own voice assistant like Alexa, Google or Siri

• After spending months doing research on how to customize alexa voice service to modify it to our own voice assistant
• We realised that these big voice service companies doesn't let you do that, simply because then no one will purchase their products.
• The only way out was to build our own Voice Service.
• This NLP solution can do weather forecast, solve small computational problems, play music, do wikipedia search and crack jokes.
• Here we are using Socket programming to establish communication between android device and the voice assistant.

Face Recognition & Temperature sensing attendance marking system

• Covid has certainly changed the world that we know, and to be in sync with the rest of the world, we should embrace this change too.
• Keeping that in mind we thought of designing an attendance system which is TOUCH FREE.
• A system that can identify someone by seeing his/her face, scan the body temperature without any touch interaction and mark his/her attendance.

Voice command based - 2D facial scanning and recognition system

• This project is implimented with the help of two Machine Learning solutions NLP (natural language processing) and Computer Vision.
• The main challenge here was to train the model in real time with very less dataset without using GPU.
• Keeping the requirements in mind, here we took the help of "Triplet loss function" rest is the coding implementation.

Android Based Endless running game

• Inspired from the games like "Subway Surfers" and "Temple Run" this is an endless runner game.
• Here the player is suppose to cover maximum distance that he could by avoiding all the obstacles on its way.
• There are also magical elements in this game where the player can reverse, freeze, teleport and move fast in Time.

Motion Tracking Gesture Control Simulation - Unity3D Project

• This is the final look of Hand Gesture controlled Simulation, i designed using Unity3D and Leap Motion as a tracking Hardware for my final year project.
• If you want to know more about this project please watch the full video.

University of Mumbai, Mumbai

Mixed Martial Arts