Arnav Garg Email and Phone Number

▪ Developed an algorithm for object detection and cell split detection in series of 3D images, targeting three types of cells within a channel.▪ Processed a dataset of approximately 650 GB, performing image segmentation to eliminate noise, ensuring the extraction of meaningful image content.▪ Utilized a Faster R-CNN with ResNet-50 model (Deep learning) as the foundational architecture, incorporating transfer learning to minimize training time and leverage pre-trained models on large… Show more ▪ Developed an algorithm for object detection and cell split detection in series of 3D images, targeting three types of cells within a channel.▪ Processed a dataset of approximately 650 GB, performing image segmentation to eliminate noise, ensuring the extraction of meaningful image content.▪ Utilized a Faster R-CNN with ResNet-50 model (Deep learning) as the foundational architecture, incorporating transfer learning to minimize training time and leverage pre-trained models on large datasets.▪ Employed a private cloud equipped with NVIDIA GPUs for training, fine-tuning and testing the model.▪ Created a custom function to track individual cells across video frames, monitor the formation of new cells, and identify their origin cells. Visualized the progression and increase of each cell type through graphical representations.▪ Also Developed a data analytics app in R using the Shiny framework. Integrated a concise report generation feature.▪ Enabled data loading in both processed RDS and raw FASTQ formats. Implemented interactive graphs using ggplot2 and Plotly.▪ Submitted the app to Bioconductor, a reputable distributor of open-source analytical software in the biological data domain. Show less

▪ Developed a Loxcode-based class hierarchy for cell classification.▪ Generated simulated data for various hierarchies due to the limited availability of real-world data, anticipating an increase in data volume in the future. ▪ Tested multiple models including KNN, Decision Tree, Random Forest, and deep learning (convolution-based time series) to determine the best fit for current and future data.▪ Reported that Random Forest produced the best results for smaller datasets, while… Show more ▪ Developed a Loxcode-based class hierarchy for cell classification.▪ Generated simulated data for various hierarchies due to the limited availability of real-world data, anticipating an increase in data volume in the future. ▪ Tested multiple models including KNN, Decision Tree, Random Forest, and deep learning (convolution-based time series) to determine the best fit for current and future data.▪ Reported that Random Forest produced the best results for smaller datasets, while convolutional model excelled with significantly larger datasets.▪ Created a pipeline to train and test the Random Forest and convolution based model with new real-world data as it becomes available.▪ Implemented new features to Data Analytics app including implementation of new Graphs. Show less

▪ Created 3 chatbot using IBM watson assistant.▪ Worked in a team of 4 to create Webapps using angular8 and node js.▪ Worked with MongoDB (noSQL database) for storage of userlogin info and dynamic data loading in webapps.▪ Implemented machine learning (ml) models in IBM watson studio and deployed these models on IBM watson machine learning.▪ Utilised the ml models in webapps using api calls.

▪ Achieved inter zabbix server communication using python script and automated the process of creating and updating a redundant server to be used to provide fault tolerance.▪ Worked in a team to automate managed IT operations through data visualization tool, kibana, and search and logging tools, ELK stack.▪ Wrote python script to Monitor and fetch data from linux and sunos servers over a ssh connection.

▪ Worked on Web Applications to meet the clients needs▪ Worked on front-end of login and registration page with Android Studio.▪ Developed communication skills required in a professional environment.