GenAI -LLM - RAG Projects

Developed biomedical GenAI agent specifically engineered to identify and validate health-related assertions. This is achieved through the use of a scraper and crawler that are generated and collects data nightly, extracting information and gathering evidence from reliable sources such as PubMed, WHO, and many more.
Utilizes LLM Ollama3.1 to produce determinations categorized as Supported, Contradicted, or Inconclusive, along with justifications and plain-language elucidations founded on scientific evidence. Outcome is the Health Information Verification System, which offers conclusions, safety guidelines, and references across multiple platforms.

Developed an AI-powered HR assistant to process and respond to queries from Nestl´e’s HR policy documents. The system was implemented in Python, incorporating PDF parsing, vector embeddings, and integration of large language models (LLMs) within a Gradio-based interface for interactive communication.
Designed the assistant to retrieve accurate, context-relevant answers directly from policy documents, with the Gradio interface enabling accessible interaction and reducing manual lookup effort.

Developed an AI-powered assistant to curate reliable, real-time news and answer general queries, addressing challenges of misinformation and fragmented news consumption.
Implemented in Python, leveraging LangChain, HuggingFace Transformers, NLTK, Pandas, and vector embeddings, integrated with news APIs and web sources (e.g., Google News, Reuters, Bloomberg), and deployed via a Gradio interface for user interaction.
Developed a working prototype that can filter misinformation, categorize news into fields such as technology, finance, and sports, and offer immediate context-sensitive responses, thereby showcasing the practicality of document- and API-based conversational systems in information management.

Created and developed a business intelligence assistant to help aimed at aiding organizations, particularly SMEs, in converting raw business data into practical insights.
Developed using Python, utilizing LangChain, Retrieval-Augmented Generation (RAG), Large Language Models (LLMs), Pandas, along with Matplotlib-Seaborn for data analysis, generating natural language insights, and presenting outcomes via clear visualizations. Produced a prototype that could analyze structured datasets, detect significant patterns, offer data-driven recommendations, and visualize insights, showcasing the capabilities of AI-powered decision support systems for business intelligence

Created a custom ChatGPT utilizing eye-tracking data. It retrieves information and can reference and cite any paper or literature associated with eye tracking. This model offers responses to all queries related to eye tracking, ensuring that every answer is supported by an appropriate citation from the literature. Serving as an eye-tracking specialist, this tool provides search capabilities for research, product development, and software applications.

Ask me for upcoming Project!

Investigated the role of fog computing in healthcare applications, focusing on real-time data processing, reduced latency, and enhanced security.
Designed a Windows-based security application using voice, face, and gesture recognition, demonstrating its feasibility in IoT environments.
Created a smart AI assistant for home automation that integrated voice, facial, and gesture recognition using Microsoft Kinect and Arduino.

Designed a sensor-equipped car seat with voice assistance to enhance mobility and safety for visually impaired individuals during vehicle travel.
Proposed an accessible airport transportation system to improve mobility, usability, and comfort for physically disabled travelers.
Investigates how middle-aged individuals interact with security and privacy settings on smartphones, identifying behavioral patterns and comprehension levels.
Conducted a qualitative analysis of privacy policies and user trust in COVID-19 testing platforms to assess data security concerns.
Explores how international students used technology (social media, apps, and virtual tools) to cope with stress and isolation during the pandemic.
Assessed the impact and effectiveness of Voice User Interfaces (VUI) in collaborative environments like classrooms and industries.
Analyzed Amazon Alexa reviews to identify potential privacy violations and security risks in third-party voice applications.
Developed a chatbot-based testing tool to evaluate security loopholes in Amazon Alexa’s skill approval process, revealing vulnerabilities in its privacy policies.

Developed an AI-powered interactive learning tool that helps users understand and explore eye-tracking technology and applications.
Description text goes here
Item description
Built a multi-stakeholder educational platform to connect teachers, students, administrators, and parents for seamless communication and learning.