Sangeeth Pogula

As a motivated and skilled Machine Learning Engineer, I bring hands-on experience gained through an intensive internship, where I successfully contributed to a range of projects involving AI and machine learning technologies. My background includes expertise in designing and developing advanced solutions that leverage machine learning models and AI techniques to address real-world challenges and enhance user experiences.

1. Ai-powered chatbot, Stringer

Project Overview: Stringer is an AI-powered chatbot designed to manage and respond to queries related to organizational policies, procedures, and guides. It aims to assist leaders by providing prompt, accurate information to ensure comprehensive understanding and compliance. Worked on this project during my internship at Brane Enterprises Pvt.Ltd, Hyderabad.

Responsibilities:

• Semantic Chunking: Implemented semantic chunking of organizational policy documents to enhance information retrieval and response accuracy.
• QA Pair Refinement: Identified and eliminated irrelevant QA pairs generated from policy documents, ensuring only high-quality question-answer pairs were used.
• Vector Databases: Experimented with and optimized vector databases (e.g., Pinecone, Milvus) for efficient storage and retrieval of high-dimensional data.
• Embedding Models: Experimented with various embedding models and their combinations with different vector databases and similarity metrics to improve the retrieval accuracy for chatbot.

2. AI-Tutor Project (Learning Platform)

Project Overview:

AI-Tutor is a comprehensive learning platform that generates video tutorials, quizzes, and FAQs from a single document containing topic-related information. The platform features a chatbot to clarify user queries related to the tutorial, leveraging large language models (LLMs) and natural language processing (NLP) techniques, specifically Retrieval-Augmented Generation (RAG). Contributed end-to-end to this project, ensuring a seamless and engaging learning experience for users. Worked on this project during my internship at Brane Enterprises Pvt.Ltd, Hyderabad.

Responsibilities:

• Video Generation: Initiated the video generation process by experimenting with various models and finalizing a face visually dubbing model to create dynamic and realistic tutorial videos. Developed a system to generate different tutorial videos dynamically from a single video source.
• LLM Experimentation: Experimented with various open-source LLMs to generate FAQs and quizzes. Utilized prompt engineering techniques to enhance the quality and relevance of the generated content.
• UI Development: Designed and developed a user-friendly video player interface using HTML, CSS, Bootstrap, and JavaScript. Implemented features including synchronized typewriting text effects and video controls. Added an in-video quiz that prompts users upon tutorial completion, and integrated FAQs with videos where an avatar speaks the answers.
• UI Hosting: Hosted the UI using Django for demo purposes, ensuring the platform's accessibility and functionality for stakeholders.
• API Development: Created APIs using FastAPI for handling video generation, chatbot interactions, FAQs, and quiz generation.
• Hardcoding and Solutions: Hardcoded all features of the project and developed innovative solutions to implement the required functionalities effectively.

3. Wild animal Intrusion detection in farmland using deep learning

Project Overview:

Our project addresses the challenge of human-wildlife conflict in agriculture using the YOLOv8 object detection model. By processing and analyzing real-time camera feeds, our system detects wild animals and provides immediate alerts through emails and buzzing sounds. This proactive approach aids farmers in taking swift action to protect their land. Completed this project as Major project as part of my academics in VNR Vignana Jyothi Institute of Engineering and Technology, Hyderabad.

Responsibilities:

• Dataset Collection and Annotation: Collected a diverse dataset featuring a wide range of animal classes in various environments and annotated it for use in model training.
• Model Fine-Tuning: Fine-tuned a pretrained YOLOv8 image detection model on the dataset to achieve high accuracy in animal detection.
• System Architecture Development: Designed the system architecture to integrate live camera feeds with the detection model. Implemented notifications via email and buzzer for detected animals.
• UI Development: Created a user-friendly interface using HTML, CSS, Bootstrap, and JS for the front end, with Flask as the backend. The UI includes admin controls for monitoring detections across multiple farms and users.

• Machine Learning Specialization | Coursera

Completed Stanford University's DeepLearning.AI Machine Learning Specialization by Andrew Ng. Mastered AI fundamentals and practical ML skills in 3 courses. Built models with NumPy, scikit-learn, and TensorFlow for classification, regression, and reinforcement learning. Applied best ML practices and explored unsupervised techniques like clustering. Developed recommender systems using collaborative filtering and deep learning methods.

• Data Science for engineers | NPTEL

Completed NPTEL's "Data Science for Engineers" by Prof. Ragunathan Rengasamy and Prof. Shankar Narasimhan, IIT Madras. Covered foundational data science concepts, algorithms, and practical R programming applications. Developed skills in data manipulation, visualization, and predictive modeling for effective data-driven solutions.

• Published a paper titled "YOLOv8-Based Wild Animal Intrusion Detection System for Farmland Protection" in The International Journal of Science, Mathematics and Technology Learning. This project utilizes advanced computer vision with YOLOv8 to enhance farmland security by detecting wild animal intrusions in real-time. Proud to contribute to wildlife conservation and agricultural protection through innovative technology. https://ijsmtl-cgrn.com/wp-content/uploads/IJS2024-31-1-23.pdf

• Member of VJ Data Questers | VNRVJIET, Hyderabad
• Member of Creative Arts Club | VNRVJIET, Hyderabad

• Coding
• Anime
• Photography
• Pencil Sketching
• Gaming