Parth Makwana

AI/ML practitioner with expertise in deep learning, computer vision, NLP, and knowledge graphs. Proficient in semantic segmentation, RAG pipelines, and workflow automation utilizing UNet, Neo4j, and LLMs. Currently pursuing opportunities in research and education, emphasizing mentorship and applied AI research.

• Neo4j professional
• Azure AI Fundamentals Certified
• Vanderbilt University certified - prompt engineer

Automated detection of coral bleaching : a deep learning approach for reef health monitoring, published in The Academic Research Journal, Volume 2, Issue 12, Dec 2024

• Developed an AI-based system using YOLOv8 and a custom coral reef dataset to detect coral bleaching with up to 94% accuracy, integrating a GUI for real-time reef health monitoring and conservation support

1. Felcita – AI-powered skin analyzer 

• Designed semantic segmentation models for identifying skin issues like pigmentation and acne, trained on custom datasets with real-time image analysis
• Trained on custom datasets, with real-time image analysis
• Facilitates dermatological assessment and cosmetic recommendations

2. Broccoli – AI-based automated bank loan approval system

• Built an AI-driven loan approval engine using Neo4j knowledge graphs and LangChain agents, modeling applicant profiles, policies, and risk rules for auto-decisions
• Improves transparency and efficiency in financial workflows

3. NLA – Natural Language to SQL Agent

• implemented a natural language interface to convert user queries into SQL using LangChain and CrewAI
  Incorporates knowledge graphs for schema understanding.
• Enables intuitive, accurate database interaction.

4. AI-Generated Audio Detection

• Built a deep learning system using CNN and LSTM to detect AI-generated audio via numeric features and spectrograms.
• Deployed with Streamlit for real-time classification and visualization.
• Aimed at media verification, cybersecurity, and forensic applications.

5. Landslide Prediction using Machine Learning

• Developed ML models (e.g., SVM, Random Forest) trained on geospatial data to predict landslide-prone regions.
• Conducted performance comparisons across six classifiers.
• Supports early warning systems and risk mitigation efforts.

6. Cartoonify an Image using OpenCV

• Created an image processing app to convert photos into cartoon-style images using edge detection and color clustering.
• Utilized OpenCV and K-Means for feature simplification.
• Interactive GUI allows upload, preview, and export.

7. Automated detection of coral bleaching using YOLOv8

• Trained YOLOv8 on underwater images to detect bleached vs healthy corals with high precision.
• Enables real-time reef monitoring and conservation support.
• Demonstrates deep learning’s impact in marine ecology.

8. Resume Shortlister – AI-Based Talent Screening Tool

• Created a chatbot-driven screening tool using Graph RAG and knowledge graphs.
• Extracts and filters resumes based on skills or experience.
• Streamlines talent acquisition with conversational UI.