• Engineered and deployed LLM, Falcon, for care plan generation, leveraging clinical notes, resulting in time reduction for care managers. NLP, LLM, LangChain, Huggingface, Streamlit, Flask.• Developed and executed Machine Learning solutions for the US healthcare industry, managing data from over 400 hospitals and achieved an impressive average prediction accuracy of 85% using algorithms such as Decision Tree, Random Forest, and XGBoost. • Optimized clinical notes processing services, reducing processing time by 40% and improving overall efficiency. • Analyzed a dataset of over 10 million health claim records, extracting actionable insights for data-driven decision-making and delivered cost reduction solutions to insurance companies, resulting in annual savings of over $500,000. • Leveraged a range of technologies, including Databricks, Machine Learning, Keras, Scikit-learn, Excel, PySpark, and Data Analysis

𝗥𝗲𝘁𝗮𝗶𝗹 𝗔𝗿𝗯𝗶𝘁𝗿𝗮𝗴𝗲 𝘂𝘀𝗶𝗻𝗴 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝗰𝗲• Integrated a retail arbitrage system by scraping real-time data from 6 major retailers every 7 minutes. • Orchestrated the integration of MongoDB for seamless data storage of the product catalog, facilitating rapid identification of price fluctuations and empowering the team to optimize pricing strategies, resulting in a 30% revenue growth. • Adopted a robust 'Name Matching Algorithm" utilizing BERT and feature matching techniques to enable informed buying and selling decisions across platforms; reduced product misidentification by 45% and increased cross-platform sales by 30%𝗞𝗲𝘆 𝗣𝗼𝗶𝗻𝘁𝘀 𝗗𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝗶𝗻 𝟯𝗗 𝘀𝗽𝗮𝗰𝗲 𝘂𝘀𝗶𝗻𝗴 𝗥𝗲𝘀-𝗡𝗲𝘁• Implemented a project to detect key points in 3-dimensional space using the Res-Net model in Python. • Detected 8 key points for each of the 6 vertebrae, encompassing Lumbar section vertebrae and Sacrum, facilitating sequential alignment of the vertebrae. • Detected a total of 132 points, representing XYZ coordinates for each point, including 40 points for each Lumbar vertebrae and 4 points for the Sacrum. • Curated and processed a dataset containing 900 marked vertebrae and 200 marked sacrum, leveraging advanced data cleaning techniques to enhance accuracy and optimize training for the machine learning model. • Accomplished exceptional performance, having an upper limit on the average error of only 5 points on the test set.

𝗦𝗽𝗶𝗻𝗲 𝗦𝗲𝗴𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 𝘂𝘀𝗶𝗻𝗴 𝗗𝗲𝗲𝗽 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 (𝗨𝗡𝗲𝘁)• Developed a product to reduce surgery time by segmentation of spine using Python, Deep Learning, and TensorFlow using the UNet model to process CT-scans, performing localization, cropping, and segmentation of the vertebrae. • The dataset included 400 sacrum segmented CT scans and 350 segmented vertebrae CT scans. • Attained outstanding performance with a dice coefficient of 0.95, ensuring precise segmentation. • Led the implementation of a cutting-edge 3D modeling system, revolutionizing the surgical planning process; facilitated precise visualization of segmented vertebrae, leading to a 40% increase in surgical accuracy and outcomes.

Learned the implementation of several Machine Learning and Deep Learning algorithms in a fast pacedorganizational environment.

• Collaborated with business school professors on a data analysis project focusing on investors' portfolio data. • Engineered a Python script to process and analyze 1M+ rows of customer data, surpassing Excel's limitations and reducing data processing time by 70%. • Developed and integrated a data-driven approach to investment portfolio analysis, leveraging programming and statistical modeling to optimize asset allocation and generate a 20% increase in portfolio performance.