Developing APIs using C# for banking solutions for the Client Coastal Community Bank, with continuous integration and deployment on Azure cloud along with CosmosDB. Developed code to deploy Azure Resources using Bicep files. Also, debug and resolve CI/CD Azure pipeline issues.

Leveraged Postman scripts and Newman for API testing automation, and prior to integration with CI/CD pipelines.

Tools used: C#, Microsoft Azure, CosmosDB, Azure Bicep, Postman, NewmanCLI.

I worked with 3D models, generating point clouds and training deep learning models, such as PointNets, to predict topological features like verts, edges, and faces for creating game asset meshes.

My responsibilities included enhancing PointNet++ architecture, training models, conducting experiments, and researching recent innovations in generative 3D AI to improve learnability using Python and PyTorch.

Developed a deep learning model with 97% accuracy in predicting 3D orientations, created a Dockerized API with the model in ONNX format for efficient CPU deployment, and benchmarked inference times on GPU and CPU.

Optimized a 3D alignment model by reducing the VGG-16 embedding size by 70% used for similarity search, and reduced processing time by 40% without changing predictive accuracy.

Tools used: Python, Pytorch, Pytorch3D, Kaolin, Numpy, Scipy, ONNX, ONNXRunTime, AWS, FastAPI, PointNets, WanDB.

Contributed to the integration of 6GHz band (Wi-Fi 6E) support in WingOS modules for Access Points and Controllers, while also resolving priority customer-reported defects.

Worked on developing a WIPS feature for terminating WPA3 connections using role-based firewalls.

Enhanced SNMP features by optimizing and implementing caches to speed up bulk response times when used with intensive network monitoring tools like Statseeker.

Supported synthetic NICs on Cloud Controllers deployed on Microsoft Azure and Hyper-V instances.

Designed and prototyped an ETL pipeline to collect network statistics from APs and transmit them to InfluxDB using Telegraf, enabling comprehensive real-time network monitoring.

Developed several features and command line functionalities for WiNG OS and also handled system maintenance.

Tools used: C, C++, Python, Wireshark, GDB, Azure, InfluxDBv2, Grafana, Make, Bash Shell.

I mentored an intern group of close to 25 students on multiple projects involving Predictive Maintenance for On-Road Wind Turbines and Energy Output Estimation based on Vision-based Vehicular Speed Prediction.

Prototyped both projects with real-time sensor data and video data for one turbine setup.

Tools used: Python, Pandas, Scikit-Learn.

Conducted research to leverage 3D Point Cloud data for advanced dimensional inspection and defect detection, focusing on overcoming occlusion and depth estimation challenges in 2D imagery.

Conceptualized and prototyped a system for precise 3D object dimension estimation using PointNet++ Part Segmentation, following extensive PointCloud data gathering, labeling, and augmentation.

Enhanced system efficiency by integrating YOLO-based object detection algorithms for high-speed quality inspections, achieving an 18% reduction in processing time.

Tools used: C++, Python, OpenGL, ONNX, Pytorch.