Passionate machine learning engineer with experience in computer vision, developing ML models and systems, data collection, exploration and analysis, enhancing existing models, deploying solutions to real world, possess good research and literature reading experience, can bring relevant ideas from state of art literature, proficient in collaborating with teams, has done some code contributions to open source projects including keras, At The Sound of AI - Open Source Research developed model that decreased detection error rates by 20%
Initiated the application of transformers in automatic speech to text module of NeuralNet music synthesizer which is an open-source research project with team of voluntary researchers and software professionals around the world, achieved decrease in error rates of output text from model by 20% by fine tuning, applying data augmentation and adding decoding algorithms to model output. - Python | Pytorch | HuggingFace Transformers | Async programming | Git | librosa
Created a cheap alert system for monkey theft detection, accomplished low error rates by using state of art detection algorithms at that time, increased detection fps from 1 to 24 by converting trained model to tensorflow lite which enabled the system to be deployed in a raspberry pi, also provided software solution for previously installed ip cameras - Python | Keras | TensorFlow | OpenCV | Multithreading | YOLO | SSD | Git | Web Scraping | Raspberry PI | TF Object Detection API
Co-Founded game and app development business, developed over 8+ games, 3+ mobile applications and 2+ softwares solutions out of which 9 apps and games were delivered and published to google play store with two apps over 40K+ and 13K+ downloads - Unity | C# | JAVA | Python | Flask | PHP | Git | Project Management | Project Design
Thesis- Application of object detection in home surveillance system Percentage- 76.55% | Division- First | Rank- 1
Implementation of style transfer for photos and videos as proposed in paper Perceptual Losses for Real-Time Style Transfer and Super-Resolution, which is 100 times faster than previous implementations. - Tensorflow | Keras | OpenCV
An object detection based surveillance application, uses live video stream from camera to detect intruders and send alerts when detected. - Python | Tensorflow | Keras | Multithreading | SSD | Git
wav2vec2 based automatic speech recognition system, takes speech signal as input and outputs text transcriptions asynchronously. - Python | Pytorch | hugging face transformers | speech recognition | Git
Javascript package for integrating WYSIWYG editor to write blogs for websites, comes with inbuilt content editing toolbar and is easy to integrate. - JavaScript
Technical Skills - Python | C# | Pandas | Scikit Learn | TensorFlow | Pytorch | Keras | Lightning | Huggingface | Linux | Machine Learning | Deep Learning | Data Analysis | Computer Vision | Signal Processing | Web Scraping | Basic NLP
Soft Skills - Project Management | Peer Tutoring | Self improvement | Collaboration | Understanding Research Papers