Harshit Joshi

Dedicated and analytical Credit Risk Analyst with 1 year of experience. Equipped with strong educational background holding masters degree in Computer Science from Indian Statistical Institute. Proficient in machine learning and data analysis, with proven sills in improving risk assessment models.

Seeking to transition into data science role to leverage and expand expertise in data driven decision-making.

• Secured AIR 30 in Indian Statistical Institute(ISI) entrance exam 2021
• Nominated for best thesis in M.Tech Computer Science
• Secured 96.41 percentile in GATE CS 2021
• Rating of 3 star at Codechef(id-harshit740)

• Data Structures and Algorithms
• Machine Learning, Neural Networks and Information Retrieval
• Statistics and Probability
• Operating Systems and Database Management Systems

Modeling earthquake richter and impacts

• Developed predictive models to assess and predict earthquake damage grades, aiming to improve disaster response and preparedness. Focused on identifying key factors such as earthquake magnitude, depth, location, and building characteristics.
• Implemented and compared the performance of various classifiers like Logistic Regression, Random Forest, SVM and LightGBM to determine the most effective model. Evaluated models based on accuracy, precision, recall, and F1-score, identifying key predictors of earthquake damage and developing a robust model to aid in disaster management and mitigation strategies.

Reconstruction Of Photoacoustic Images Using Sparse 

• Enhanced the quality of photoacoustic images by developing deep learning techniques to remove artifacts caused by sparse data acquisition, improving medical diagnostic accuracy.
• Developed and implemented a Modified U-Net and a Modified GAN architecture, achieving promising results in artifact removal and image reconstruction compared to state-of-the-art algorithms.

Personalized Email Response System 

• Developed a deep learning model to automate personalized email responses, enhancing customer service efficiency.
• Developed personalized,context-aware responses using advanced NLP.
• Fine-tuned a pre-trained T5 model using the Hugging Face Transformers library ,configured and optimized training parameters for improved model performance.