Machine Learning / Data Scientist

About the Project We are looking for an experienced Machine Learning Engineer / Data Scientist to design, develop, validate, and deliver a production-ready Lead Scoring / Propensity-to-Convert Model. We will provide the required historical data. The objective is to develop a robust predictive model that assigns a propensity score to each lead, allowing sales and marketing teams to identify and prioritize leads with the highest probability of conversion. We are particularly interested in candidates with prior experience developing lead scoring, propensity, conversion prediction, customer analytics, or sales optimization models. This is a fixed-price, deliverable-based project. Successful completion may lead to additional AI/ML and Data Science projects. Scope of Work The selected candidate will be responsible for the complete model-development lifecycle, including: Understand the supplied datasets, target variable, business objective, and conversion definition. Perform data quality assessment and exploratory data analysis (EDA). Identify missing values, outliers, class imbalance, data leakage, duplicates, and other data-quality issues. Perform appropriate data preprocessing and transformation. Conduct feature engineering and feature selection to identify meaningful predictors of lead conversion. Establish an appropriate baseline model. Develop and compare suitable machine-learning approaches such as: Logistic Regression Random Forest XGBoost/LightGBM or similar gradient-boosting approaches Other appropriate ML techniques where justified Perform hyperparameter tuning and model optimization. Address class imbalance appropriately where required. Generate a propensity/conversion probability score for each lead. Define meaningful lead-score bands/segments such as High, Medium, and Low Propensity, based on model performance and business usefulness. Evaluate the model using appropriate metrics such as ROC-AUC, Precision, Recall, F1, PR-AUC, Lift/Gain, calibration and confusion matrix, as applicable. Provide feature importance/model explainability, preferably using SHAP or an equivalent technique. Validate model stability and generalization on unseen data. Recommend an appropriate score threshold or prioritization strategy based on business objectives. Produce clean, reusable, well-documented Python code. Provide complete model documentation and knowledge transfer to our internal team. Expected Deliverables At the completion of the project, we expect: EDA & Data Quality Report Feature Engineering/Selection Documentation Baseline and Candidate Model Comparison Final Optimized Lead Scoring Model Lead-Level Propensity Scores / Conversion Probabilities Lead Segmentation/Ranking Methodology Model Performance & Validation Report Feature Importance / Explainability Analysis Model Artifact and Inference/Scoring Code Clean, documented Python source code/notebooks README / Technical Documentation explaining how to reproduce, retrain, and score new data Final Presentation and Knowledge-Transfer Session with our team The final solution should be reproducible and sufficiently documented for our internal technical team to maintain and extend. Required Skills We are looking for candidates with strong hands-on experience in: Python • Machine Learning • Data Science • Predictive Modeling • Lead Scoring • Propensity Modeling • Classification • Feature Engineering • Scikit-learn • Pandas • NumPy • XGBoost/LightGBM • Model Evaluation • SHAP/Explainable AI • Statistical Analysis Experience with customer conversion models, CRM/sales data, marketing analytics, automotive analytics, or high-volume consumer lead data is highly desirable. Ideal Candidate You should have demonstrated experience taking ML projects from raw data through validated model delivery, rather than only building experimental notebooks. You should be comfortable explaining: Why you selected a particular modeling approach. How you prevented target/data leakage. How you handled imbalanced conversion data. How you evaluated whether the model actually improves lead prioritization. How your propensity scores can be used operationally by sales teams. How the model should be monitored and retrained over time. Confidentiality & Intellectual Property The selected freelancer will be required to maintain strict confidentiality regarding all datasets, business information, methodologies, and project materials provided. All project-specific code, trained model artifacts, documentation, analysis, and deliverables created and paid for under this engagement must be provided to us in accordance with the agreed contract terms. No project data may be uploaded to public repositories or shared with third parties. When Applying Please answer the following questions in your proposal: Have you previously built a lead scoring, propensity-to-buy, conversion prediction, or similar classification model? Briefly describe one example. What algorithms would you initially consider for a lead-scoring problem, and why? How would you evaluate a lead-scoring model when the conversion rate is highly imbalanced? How would you detect and prevent data leakage? Which metrics would you use to demonstrate that the model helps sales teams identify better leads? How would you convert model probabilities into actionable High/Medium/Low lead segments? What tools/libraries would you use? Please provide an example of a similar project, GitHub repository, portfolio, or anonymized model-performance report if available. What is your fixed-price quote for completing the entire project? What is your estimated timeline? Please begin your proposal with the words “Lead Scoring ML” so we know you have read the complete project description.