Chayma Rhaiem
Data Scientist & Software Engineer
Equipping LLMs with facts, features, and a side of ✨delusion control✨
Software engineer focused on data science and applied AI. I build systems that combine machine learning, cloud infrastructure, and retrieval-augmented generation (RAG) to solve practical problems
About Me
Data Scientist and Software Engineer with expertise in machine learning, cloud computing, and intelligent system design.
AI/ML Engineering
RAG systems, NLP pipelines, and scalable ML architectures
Data Engineering
Cloud data pipelines, vector databases, and real-time processing
Software Engineering
Full-stack development, system design, and production deployment
Professional Experience
My journey in data science and software engineering
Data Scientist
Avaxia Group
- Oct 2024 - Jun 2025: Designed hybrid Graph-Vector RAG system integrating Graph and Vector cloud data storages for SAP anomaly detection
- Built automated recommendation engine with knowledge graphs for real-time fix suggestions, scaling alert monitoring to process 70x more data points
- Initial Project (Jul - Sep 2024): Developed Microsoft Teams Alerting Chatbot cutting SAP anomaly detection time, using generative AI and vector RAG
- Streamlined data pipelines in AWS and SAC, improving data processing efficiency by 30%
- Delivered comprehensive reports and interactive dashboards, reducing data retrieval time by 40%
- Led development of AI-based chat assistant PoC for major gas company using AWS services and LLMs
Key Projects
AWS Glue to Power BI Pipeline
Built comprehensive ETL pipeline using AWS Glue for data transformation and Power BI integration, enabling real-time dashboard analytics for enterprise clients with automated data quality checks.
Bedrock Streamlit Chatbot with Claude Foundation Model
Pioneered early RAG implementation using AWS Bedrock's foundation models for intelligent document processing. Built when RAG was just emerging as a paradigm, featuring advanced prompt engineering and context-aware responses.
AWS S3 and Redshift Data Pipeline Optimization in Glue
Developed proof-of-concept AI assistant for major gas company operations, integrating multiple AWS services for document analysis, query processing, and automated response generation.
- Leveraged datasets from BFI's African banking partners (300,000+ customer records) for credit risk modeling
- Refined credit risk models under Basel II and IFRS 9 guidelines improving prediction accuracy
- Enhanced prediction accuracy by 10% through advanced model optimization and AutoML techniques
Education
Academic foundation in software engineering and data science
Bachelor of Engineering in Software Engineering - Data Science Specialization
ESPRIT (Private Higher School of Engineering and Technology)
Key Achievements
- Engineered intelligent project management assistant using Graph RAG architecture with LLAMA models
- Developed intelligent trading agent for Tunisian financial market using ML forecasting
Relevant Coursework
Bachelor of Science in Computer Science - Software Engineering and Information Systems
Polytech-Intl
Key Achievements
- Created automated Instagram Stories Scraper with Google Cloud Platform
- Implemented computer vision models for brand detection
Relevant Coursework
Certifications
Professional certifications in data science, cloud computing, and machine learning
Microsoft Certified: Azure Data Scientist Associate
Microsoft
SAP Certified Application Associate - SAP Analytics Cloud
SAP
Natural Language Processing Specialization
DeepLearning.AI
Building Transformer-Based NLP Applications
NVIDIA
Google Data Analytics Specialization
Attendance Hashgraph Developer
The Hashgraph Association
Technical Skills
Expertise across the full data science and software engineering stack
Machine Learning
Programming
Web Scraping & Tools
Cloud & Data
Languages
Featured Projects
Academic research and professional implementations
Highlighted Professional Projects
Hybrid Graph-Vector RAG System
Hybrid RAG architecture combining graph and vector retrieval, Custom G-Eval framework for LLM response quality assessment , 70% alert accuracy matching expert-defined priorities, Conversational interface for natural language SAP queries, Automated RCA (root cause analysis) with contextual documentation
Credit Risk Assessment System
Automated credit PD (Probability of default) models under Basel II and IFRS 9 guidelines using 300,000+ customer records from African banks.
Highlighted Academic Projects
Object Detection Using Adaptive Mask RCNN in Optical Remote Sensing Images
This project presents a comparative study of three deep learning approaches for classifying images in the CIFAR-10 dataset: A basic Artificial Neural Network (ANN), A custom Convolutional Neural Network (CNN), Transfer Learning using VGG19 and ResNet-50.