sandesh athawale

Utilized PTZ camera zoom & pan capabilities to achieve high-precision pest detection from a distance. Integrated OpenCV and Flask to visualize pest locations on a web application.

Developed AI models in Python for 3D point cloud semantic segmentation, enabling automatic identification of structural elements like walls, ceilings, and floors. Deployed the solution using AWS (EC2, S3) with a Flask backend for efficient processing and seamless integration.

Developed a chatbot leveraging OpenAI GPT-based models to retrieve and summarize information from extensive company documents, integrated with AWS (CloudFront, Amplify, API Gateway, Lambda) and Azure OpenAI Services. Optimized data management using DynamoDB and implemented streamlined information retrieval for complex documentation, reducing search time and enhancing comprehension.

Developed an object detection system using YOLO to automatically identify structural elements in floor plan diagrams. Optimized inference speed by deploying the model with TensorRT and integrating it into a web-based visualization.

Designed a modular architecture to connect the RAG system to LLMs for seamless generation of outputs. Fine-tuned LLMs using QLoRA, optimizing memory and efficiency, while improving task-specific performance.

Developed custom GAN architectures using PyTorch, OpenCV, and Flask, enabling automated generation of floor plans based on area layouts and room connectivity diagrams. Deployed the solution on AWS (EC2, S3) and Azure.

Developed a secure tree-diagram system to manage and optimize room access within a company building, utilizing AWS (EC2, S3), Flask, PyTorch, and Graphviz for backend implementation. Designed an interactive web interface with CSS, HTML, and JavaScript, offering intuitive visualization and seamless navigation of secure and restricted areas.

Independently implemented machine learning algorithms focusing on efficiency and accuracy. Solved L1/L2 problems for 5G systems using advanced AI/ML techniques.

Provided one-on-one tutoring on topics like Python, C, C++, Machine Learning, Data Science, and AI, ensuring students gained a comprehensive understanding of concepts. Addressed diverse and challenging problems daily, delivering optimized solutions and clear explanations.

Analyzed customer behavior using interaction time and visit frequency, implementing a Decision Tree algorithm on a dataset of 1000 unique customers, achieving an initial accuracy of 60%. Enhanced accuracy to 80% by applying feature engineering and switching to the CatBoost algorithm, leveraging its superior handling of categorical data and overfitting prevention.

Developed a recommendation engine tailored for vintage clothing by web scraping a dataset of 500 images, categorizing them by events and weather conditions. Trained a VGG-19 model using Transfer Learning, achieving a processing time of 1.2 seconds and an accuracy of 93%.

Grade: 7.32

Grade: 9.4/10

Completed Class X under the CBSE curriculum, achieving a commendable grade. This early education provided a solid academic base and fostered a strong work ethic.

Grade: 86.6%

Achieved a strong academic performance in Class XII under the CBSE curriculum, demonstrating proficiency across various subjects. This foundational education prepared me for higher studies and specialized technical fields.