• Collaborated closely with clients to comprehensively understand their business requirements in the realm of cybersecurity.• Spearheaded the development of an Network Security and Endpoint security using advanced Machine Learning techniques to fortify the company's digital infrastructure.• Demonstrated expertise in identifying and categorizing various types of cyber-attacks, including PortScan, DNS tunneling, ransomware, and more. Proficiently worked with Sysmon logs, Windows event logs, and packet flow capture. Achieved accuracy more then 98% on real time data as part of model performance.• Performed extensive data generation by simulating cyber-attacks to produce realistic datasets used for training Machine Learning models. This included synthesizing data on various attack types, which helped with an additional 2-3% more accuracy. Employed rigorous data preprocessing and feature selection techniques to prepare clean, high-quality datasets for model generation.• Managed complex data flows, involving fetching data from NVDAPI, storing it efficiently in a MongoDB database, transforming and formatting it into CSV, and ultimately loading it into a PostgreSQL database to serve as training data.• Leveraged Natural Language Processing (NLP) techniques and the FastText model to process textual descriptions and integrate them into the Intrusion Detection System.Technology: Machine Learning, Data Analysis, Python, Text processing, Airflow, Docker, MLFlow, Git, Jira, Vault, Prometheus, Postgres, MongoDB, CICD

• Discussed with stakeholders to define the metrics, data and scope of this project.• Initiated and completed an end to end process of labelling and anonymizing 10,000+ images to meet object detection with 96% accuracy using Python and Deep Learning.• Improved quality of images for computer vision tasks from 70% accuracy to 85% accuracy on a dataset of 10k images by applying image pre-processing techniques such as histogram equalization, barrel distortion and demosaicing.•  Utilized advanced python-based libraries, including Keras and TensorFlow to develop and integrate deep learning models into applications.• Improved model performance to 96% accuracy by tuning hyper-parameters such as epochs, dropout rate, learning rate, batch size and CV methods. Evaluated the model performance by forming a confusion matrix and deriving precision and recall using sklearn package.• Deployment on High Performance Computer (HPC) using the Docker toolkit.Technology: Deep Learning, OpenCV, LabelMG, Machine Learning, Python, SQLite3, Docker, Github.• Performed data analysis to identify patterns and aggregate data to provide insights on the cost of goods sold, reoccurring patterns in past sales, and the revenue of the company in order to improve the design and the performance of the overall data flow.• Developed Flask UI for monitoring jobs and a shell script to automate tasks, which helped in improving performance by 30 percent.• Developing and testing new functionality in existing applications for faster delivery with minimum production issues.• Implemented processes of release management, automated code deployment, configuration management and monitoring. Which results in a work efficiency improvement of 10-15 percent.• Implemented process improvements to content deliverables.Technology: HTML, Postman, GIT, Jenkins, Shell Script, IBM DataStage, IBM Data Studio, Jira, Python, Excel, Putty, WinSCP, RFHutil C.