Mohamed Jihed Riahi Data Scientist Azure Certified

Mission:

• Developed a Q&A assistant that simplifies business creation by answering our clients’ business inquiries and suggesting their next steps.
• Built a RAG-based automated reporting system using Infor CloudSuite ERP data, reducing manual report generation time by 90% (from 2 days to ~2 hours) and eliminating human data-entry errors
• Conducted a comparative analysis of Mistral and Phi vs OpenAI models through finetuning for both classification and instruction tasks, optimizing the trade-off between inference cost and accuracy compared to paid models
• Developed a Chrome extension with an intelligent chat interface that dynamically analyzes and interacts with any active tab’s content,enabling optimized browsing and instant access to information.
Stack: Python, LangChain, OCR, Prompt engineering, FAISS, Quantization, Weights and Biases, HuggingFace, Express, NodeJS

Mission:

• Built an ARIMA model to forecast revenue, improving financial projection accuracy by 15% and enabling data-driven budget allocation.
• Designed benchmark-based ELT data pipelines to consolidate operational metrics, ensuring data availability for executive dashboards.
• Developed interactive Power BI dashboards to visualize KPIs, enabling executives to track trends and adjust pricing strategies in real-time.
Stack: Statistical modeling · Microsoft Azure · Python · LangChain · Power BI · SQL etc

Mission:

[Applied Deep Learning on Black-Scholes Model]: Developed a neural network model using PyTorch to recreate the Black-Scholes option pricing model, achieving more accurate predictions by overcoming constant volatility assumptions and skewness limitations.