Developed a scalable computer vision system to classify kidney CT scans using a VGG16 architecture with MLFlow and DVC.
Deployed on Amazon ECR by using docker and EC2 instance.
A question answering LLM system capable of information retrieval from PDF files uploaded by user. The application utilizes Open AI API and FAISS vectorstore.
A Logistic Regression model trained to classify a cancer cell as benign or malignant type, deployed on Streamlit platform.
Created a interactive dashboard to change cell nuclei measurements and see how the classification results vary.s
Implemented a predictive monitoring system using trace signals and metrology data, collected from an etching product. Utilized self-organizing map (SOM) for health assessment, and Minimum Quantization Error (MQE) to calculate deviation from baseline state.
A suport vector machine (SVM) algorithm was deployed to assess the extent of degradation of the tool.
Compared various algorithms to choose the best in sklearn library to conduct Exploratory data analysis on students' test scores.
The project studied how test scores are affected by other variables such as test preparation time, gender, ethnicity, parental level of education etc.
A computer vision system that performs end-to-end pad detection from consumer images.
Various image pre-processing techniques are applied along with a trained Yolo v7 model and deployed in Azure Databricks.
Project link restricted due to confidentiality agreement with P&G.