I’ve worked extensively with structured business data sources such as Salesforce, SharePoint, internal databases, and email systems, developing classification, clustering, and predictive models to support data-driven decision-making. My focus has been on Gradient Boosting, XGBoost, and LightGBM models, enhanced with calibration techniques and SHAP-based explainability for transparent and reliable predictions.
I’ve also built FastAPI-based prediction pipelines to operationalise machine learning models, incorporating comprehensive feature engineering, encoding methods, and skewness correction. On the unsupervised side, I’ve designed clustering systems using PCA and Yeo-Johnson transformations, evaluating performance through Silhouette, Davies–Bouldin, and Calinski–Harabasz indices to uncover meaningful customer or opportunity segments.
I have also built a full-stack predictive maintenance pipeline project, which predicts machine failure before it occurs and suggests a planned maintenance procedure to reduce downtime and prevent critical machine breakdown
I’m passionate about connecting data, automation, and business strategy, creating scalable, interpretable, and high-impact machine learning systems that drive measurable results.
Please find my CV attached below
Note: I have a transferable Iqama
--- CV ---
MOHAMMED MUSTAFA
AHMED
D A T A S C I E N T I S T | A I E N G I N E E R
Email: Phone: Address:
mustafa12042002 0595874698 Riyadh, Saudi
@gmail.com Arabia
PROFILE
Data Scientist & AI Engineer specializing in end-to-end ML systems and predictive maintenance solutions.
Experienced in building production-grade models (XGBoost, LightGBM, Gradient Boosting), real-time IoT pipelines,
explainable AI (SHAP), and full-stack deployments using FastAPI, AWS, MQTT, and LLM-powered insights to drive
operational intelligence and automation.
EXPERIENCE
AI ENGINEER
Nommas.ai Riyadh, Saudi Arabia
Aug 2025 - Jan 2026
Implemented an unsupervised learning approach to establish baseline operating conditions for industrial
equipment, enabling early detection of abnormal behaviour without reliance on labelled failure data.
Integrated a Retrieval-Augmented Generation (RAG) framework to ingest and reference machine documentation,
enhancing the contextual understanding of sensor anomalies and operational patterns.
Designed adaptive, machine-specific thresholds to account for variability across assets, improving the accuracy of
anomaly detection and reducing false positives.
Leveraged large language models to translate complex analytical outputs into clear, human-readable insights,
supporting faster interpretation and decision-making by operations and IT stakeholders.
Achieved a 55% reduction in equipment downtime and proactively prevented critical machinery failures through
AI-driven predictive maintenance in a manufacturing environment.
DATA SCIENTIST
Aldrich Research services Hyderabad, India
Feb 2025 - Jul 2025
Built a revenue prediction model to estimate conversion probability and expected deal value, enabling data-driven
prioritization of sales outreach across 10,000+ prospects.
Developed and deployed a customer churn prediction system using supervised learning and LLM-powered
explainability, automatically flagging at-risk customers and generating human-readable churn insights to support
retention teams.
Designed an Ideal Customer Profile (ICP) scoring model using behavioral and firmographic features to rank high-
value prospects, contributing to a 20% reduction in customer churn through improved targeting and engagement.
EDUCATION
BACHELORS IN COMPUTER SCIENCE ENGINEERING - CSE 2020- 2024
Osmania University Hyderbad, India
SKILLS
Python Programming Feature Engineering Langchain
SQL- Basic Model Evaluation Predictive Maintenance
Data Preprocessing RAG Gen-AI
PROJECTS
Sentiment Analysis using DLT & ML
Built sentiment analysis models leveraging both traditional Machine Learning (Logistic Regression, SVM) and
Deep Learning (LSTM, CNN) to classify customer reviews with high accuracy.
Big Sales Price Prediction
Developed a predictive model using Logistic Regression to forecast high-value sales outcomes, leveraging
customer and transactional data.
Stock Index Price Prediction
Implemented a Long Short-Term Memory (LSTM) deep learning model to predict stock index price trends using
time-series data.
TECHNICAL TOOLS & LIBRARIES
Python (Pandas, NumPy, Scikit-learn, ADDITIONAL
XGBoost, LightGBM) INFORMATION
Data Visualisation (Matplotlib, Seaborn,
Plotly)
SQL for Data Extraction & Analysis
Jupyter Notebook for Prototyping & Languages
Reporting
English
Tensorflow
Arabic
Langchain & Langraph
SHAP for model assessment
Iqama Status
Retrieval Augmented Generation (RAG) Transferable
Predictive Maintenance
Front-end (Streamlit)