An experienced machine learning engineer with 4.4 years of experience using statistical modeling, natural language processing, deep learning algorithms to solve challenging business problems.
Currently working on multiple solutions that include predictive analytics, Natural Language Processing, computer vision, Speech processing etc. I have keen interest in deep learning research and currently enrolled in Fast AI course.
Classify Bear images using RESNET architecture. The classifier is trained with bear, teddy and grizzly images downloaded from google using Javascript.
Then trained RESNET classifier by finetuning fastai pretrained model. Then model artifact is stored as exported as a downloadable link using google drive.
Then the model inference code is written using Starlette framework in python and deployed in Render for free.
Personal project for Fast ai. Technologies - Python, fastai, Pytorch.
Recommend Knowledge bytes to employees based on their read history to enhance the reading experience.Recommend article to users using Hybrid Recommendation System with content and Item-Item collaborative filerting.
Increased 30% traffic for Knowledge Bytes.
Recommend similar articles based on similar topics using Latent Dirichlet Allocation algorithm. This project is made by creating a topic model on wikipedia corpus. Text preprocessing is used to clean and create a gensim model for LDA and similar articles are grouped together with Jenson-Shanon Divergence measure.
Close Project
Automate the process of vehicle damage claim assessment by using IoT drone to navigate and collect the image, process the image using CNN models and create incidents/claim cases in Microsoft D365 CRM. The SSD pretrained Object Detection model identifies the car in the scene and VGG16 image classifier categorizes the severity of the damage. IoT drone is controlled through MAVProxy and service is created using a Flask framework. Also the model identifies the license plate number using separate Open ALPR implementation.
Close Project
Lorem ipsum dolor sit amet, consectetur adipisicing elit. Mollitia neque assumenda ipsam nihil, molestias magnam, recusandae quos quis inventore quisquam velit asperiores, vitae? Reprehenderit soluta, eos quod consequuntur itaque. Nam.
Close Project
Lorem ipsum dolor sit amet, consectetur adipisicing elit. Mollitia neque assumenda ipsam nihil, molestias magnam, recusandae quos quis inventore quisquam velit asperiores, vitae? Reprehenderit soluta, eos quod consequuntur itaque. Nam.
Close Project