SRIRAM RAMPELLI

Publications

• Rampelli, S. (2025). Empowering Healthcare Data Systems With an Innovative Chatbot Application Utilizing Python and Advanced Generative AI Models. IEEE CCWC 2025.

Presented at IEEE Computing and Communication Workshop and Conference (CCWC) 2025, highlighting innovative AI-driven solutions for healthcare data systems

Projects

Empowering Healthcare Data Systems with an Innovative Chatbot Application

• Developed an innovative healthcare chatbot application that leverages advanced Generative AI models, including fine-tuned LLaMA and enhanced BioBERT, to address challenges in healthcare data accessibility, patient engagement, and operational efficiency. The system integrates state-of-the-art NLP techniques, modular architecture, and ethical AI practices to deliver scalable, real-world solutions.
  
  Key Achievements:
  
  Medical Query Resolution: Fine-tuned LLaMA using 150,000 curated medical Q&A pairs, achieving an 88% Exact Match Score (EM) and 91% F1-Score.
  Prescription Analysis: Enhanced BioBERT with custom NER layers for extracting medications, dosages, and patient details from prescriptions using OCR and advanced preprocessing techniques.
  Document and Image Analysis: Supported multi-format data, including PDFs, DICOM files, and handwritten medical prescriptions.
  Personalized Recommendations: Provided condition-specific recommendations and insights, improving patient interaction and satisfaction.
  Technical Highlights:
  
  Achieved a response time of 1.2 seconds, significantly faster than existing solutions like MedChat and Dr. AI.
  Designed a scalable architecture using Flask, Docker, and Kubernetes for asynchronous processing and optimized model serving.
  Integrated ethical considerations, ensuring compliance with HIPAA and GDPR.
  Tools & Technologies: Python, Flask, PyTorch, Hugging Face, OpenCV, Pydicom, Docker, Kubernetes, TensorFlow Serving.
  
  Impact: Improved user satisfaction with a score of 4.6/5 in studies with healthcare professionals and patients. Demonstrated potential for deployment in real-world healthcare environments by reducing operational inefficiencies and enhancing patient care.
  
  Presentation: Presented at the Computing and Communication Workshop and Conference (CCWC) 2025, highlighting its innovative contributions to healthcare AI.Developed an innovative healthcare chatbot application that leverages advanced Generative AI models, including fine-tuned LLaMA and enhanced BioBERT, to address challenges in healthcare data accessibility, patient engagement, and operational efficiency. The system integrates state-of-the-art NLP techniques, modular architecture, and ethical AI practices to deliver scalable, real-world solutions. Key Achievements: Medical Query Resolution: Fine-tuned LLaMA using 150,000 curated medical Q&A pairs, achieving an 88% Exact Match Score (EM) and 91% F1-Score. Prescription Analysis: Enhanced BioBERT with custom NER layers for extracting medications, dosages, and patient details from prescriptions using OCR and advanced preprocessing techniques. Document and Image Analysis: Supported multi-format data, including PDFs, DICOM files, and handwritten medical prescriptions. Personalized Recommendations: Provided condition-specific recommendations and insights, improving patient interaction and satisfaction. Technical Highlights: Achieved a response time of 1.2 seconds, significantly faster than existing solutions like MedChat and Dr. AI. Designed a scalable architecture using Flask, Docker, and Kubernetes for asynchronous processing and optimized model serving. Integrated ethical considerations, ensuring compliance with HIPAA and GDPR. Tools & Technologies: Python, Flask, PyTorch, Hugging Face, OpenCV, Pydicom, Docker, Kubernetes, TensorFlow Serving. Impact: Improved user satisfaction with a score of 4.6/5 in studies with healthcare professionals and patients. Demonstrated potential for deployment in real-world healthcare environments by reducing operational inefficiencies and enhancing patient care. Presentation: Presented at the IEEE Consumer Communications & Networking Conference (CCNC) 2025, highlighting its innovative contributions to healthcare AI.
• Skills: Generative AI · LLaMA · bioBERT · Fine Tuning · Optical Character Recognition (OCR) · Text-to-Speech · Data Visualization · Web Scraping · Prescriptive Analytics

Development of Multifaceted Chatbot Service Application

• Developed a comprehensive chatbot application integrating document analysis, web scraping, text generation, and voice interaction using advanced AI models and Python libraries, Implemented tools for reading and analyzing PDFs and DOCX files, performing sentiment analysis, summarization, word cloud generation, and named entity recognition (NER).
• Utilized BeautifulSoup for web scraping, transformers for summarization, VaderSentiment for sentiment analysis, and word cloud generation
• Enhanced with a fine-tuned GPT-2 model trained on Wikipedia data for improved text generation,Integrated GPT-2 for dynamic text creation and question-answering functionalities, Enabled voice commands and responses using SpeechRecognition and pyttsx3.
• Libraries Used: Flask, PyPDF2, python-docx, TextBlob, VaderSentiment, transformers, WordCloud, BeautifulSoup, GPT-2, SpeechRecognition, pyttsx3.

Chatbot Creation Using NLTK

• Leveraged the Natural Language Toolkit (NLTK), a pivotal tool in natural language processing, to develop an advanced chatbot. This chatbot is designed to deliver detailed and informative responses, showcasing my ability to apply complex NLP techniques to improve user interaction and automate customer service operations.

OpenAI Powered Poetry Application

• Conceived and developed a web-based poetry application aimed at enriching user engagement through the exploration and creation of poetry. This project involved daily challenges and features for automated poetry generation and detailed analysis, enhancing the user’s literary appreciation and creative expression.
• The application's backend was developed using Python and Flask, integrating NLTK for processing natural language and utilizing OpenAI’s API to enable dynamic content generation. This project exemplifies my capability to combine multiple technologies to deliver a seamless and innovative user experience.
• Incorporated a suite of Flask extensions including Flask-Login for robust user authentication, Flask-SQLAlchemy for efficient database interactions, and WTForms for secure and flexible form handling, ensuring the application's reliability and scalability.

Oracle Cloud Infrastructure 2024 Generative AI Certified Professional