Dens Mathew
Artificial Intelligence Specialist
Python
SQL
Wolfram
Data Pipeline & Feature Engineering: 4/10
Active 9 hours ago
+33 (7) 58817966 Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
ML practitioner (early-career, middle tier) focused on applied classification and EDA in finance and insurance with a strength in end-to-end supervised-modeling pipelines. The strongest proven skill is end-to-end model development and hyperparameter tuning as shown in DensMathew/DensMathew/insurance_fraud_detection.ipynb where GridSearchCV and an evaluation pipeline are implemented. Limited evidence of production engineering, infrastructure (CI/CD, data versioning, requirements), and advanced statistical rigor is present in public artifacts.
Technical skills
Python• Middle
SQL
Wolfram
Databases
SQLite
AI/ML
Data Augmentation
Google Colab
NumPy
Pandas
Scikit-learn
Keras
TF-Keras
Jupyter Notebook
DevOps
Azure
Analytics
Matplotlib
Seaborn
Plotly
Power BI
Tableau
Design
Adobe Photoshop
Timeline
Montpellier Business School (Ecole supérieure de commerce de Montpellier)
Master's Degree •
Big Data and AI
Market Analysis Intern
•
Junior
Brandsmith360
•
Internship
Conducted market research and competitor analysis across multiple segments to identify emerging industry trends. Produced reports and presentations summarizing consumer behavior and competitive positioning. Used Power BI and spreadsheet-based analysis to visualize insights for stakeholder review. Coordinated with team members to translate research findings into actionable business strategy.
Power BI
AI Intern
•
Junior
GAO Tek Inc
•
Internship
Built a data pipeline to clean, preprocess, and impute missing values in large business datasets using generative-AI techniques. Developed interactive dashboards to visualize key performance metrics for multiple departments. Applied AI-driven data augmentation to improve analytics readiness and support downstream modeling needs. Converted KPI requirements into analytical solutions and recommendations for stakeholders.
Power BIsince 2024
Tableau
Kannur University
Bachelor's Degree •
Bachelor of Commerce with Computer Applications
Middle AI/ML Engineer
Confidence: Medium Data-centric
Data-centric ML practitioner at a middle level specializing in end-to-end tabular data workflows and applied classification problems. The strongest proven skill is building standard tabular ML pipelines including data cleaning, feature engineering, hyperparameter tuning and evaluation as shown by the insurance_fraud_detection.ipynb clean_data, feature engineering and GridSearchCV blocks. Public code does not evidence production-grade MLOps, experiment tracking, rigorous testing, or research-level model development.
Model Architecture & Training
3/10
How well models are designed and trained
Basic model-building and training practices are present: Keras Sequential networks and standard sklearn classifiers with GridSearchCV are used, but there are no custom architectures, advanced training schedules, or rigorous training engineering.
Data Pipeline & Feature Engineering
4/10
How data is prepared for models
Clear data-centric work with explicit cleaning, outlier handling, imputation, one-hot encoding and scaling; feature engineering is applied but pipelines are ad hoc (not modularized into reusable ETL components).
Evidence
DensMathew/DensMathew/insurance_fraud_detection.ipynb:clean_data function
DensMathew/DensMathew/insurance_fraud_detection.ipynb:feature engineering block with OneHotEncoder and StandardScaler
DensMathew/K-1o-filling-Data-Analysis/BCG_GenAI.ipynb:Revenue Growth (%) and Net Income Growth (%) groupby computations
Experimentation & Evaluation
3/10
How results are measured and tested
Reasonable experimentation and evaluation practices: train/validation/test splits, GridSearchCV hyperparameter tuning, standard metrics and visualizations are used, but there is no experiment tracking, reproducible run configuration or ablation analysis.
Evidence
DensMathew/DensMathew/insurance_fraud_detection.ipynb:GridSearchCV and cv=5 usage
DensMathew/DensMathew/insurance_fraud_detection.ipynb:evaluate_model function with metrics and confusion matrix
DensMathew/DensMathew/Credit Card Client Classification.ipynb:history plotting and ROC/metrics in train_evaluvate_model
MLOps & Deployment
1/10
How models are shipped to production
Minimal MLOps or deployment artifacts: model saving is present but there is no serving, versioning, CI/CD, monitoring or containerization evidence.
Computational Efficiency
1/10
How efficiently computing resources are used
No demonstrated compute-efficiency engineering: no GPU profiling, quantization, batching optimizations beyond default settings, or memory/throughput measurements.
Research Depth & Innovation
1/10
Depth of research and new ideas
No research-level contributions or novel algorithms are present; code uses standard models and textbook pipelines without custom layers or reproduced paper implementations.
Expertise
Finance & FinTech AI• Middle
Industries
Data & Analytics• Middle
Financial Services• Middle
Technologies
SQL
Wolfram
Azure
Google Colab
SQLite
Data Augmentation
Recommendations
- Develop prototypes for tabular ML problems such as fraud detection or credit scoring where the focus is on feature engineering, model selection and evaluation.
- Build end-to-end reproducible pipelines by modularizing preprocessing into reusable ETL scripts and adding experiment tracking (W&B or MLflow) and configuration files.
- Harden projects for production: add model versioning, simple serving (e.g., FastAPI/BentoML), unit tests for preprocessing steps and CI automation.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Middle Data Scientist
Confidence: Medium ML Practitioner
ML practitioner (early-career, middle tier) focused on applied classification and EDA in finance and insurance with a strength in end-to-end supervised-modeling pipelines. The strongest proven skill is end-to-end model development and hyperparameter tuning as shown in DensMathew/DensMathew/insurance_fraud_detection.ipynb where GridSearchCV and an evaluation pipeline are implemented. Limited evidence of production engineering, infrastructure (CI/CD, data versioning, requirements), and advanced statistical rigor is present in public artifacts.
Statistical Rigor
2/10
Correct use of statistics
Basic exploratory statistics and correlation checks are present but there is little formal statistical testing, uncertainty quantification, multiple-comparison control, or causal reasoning.
Data Wrangling & Cleaning
4/10
Preparing and cleaning data
Solid, conventional cleaning and preprocessing are implemented (imputation, IQR clipping, encoding, scaling), but steps are basic and there is limited provenance, validation, or leakage checks.
Exploratory Analysis & Visualization
4/10
Exploring and visualizing data
Good variety of plots and visual checks (heatmaps, distributions, boxplots, scatter) and short written summaries, but explanations are high-level and not tied to rigorous statistical interpretation.
Predictive Modeling
4/10
Building models that predict
Complete predictive workflow exists (feature engineering, GridSearchCV hyperparameter tuning, evaluation metrics, and a TensorFlow model), but model-validation choices and error analysis are basic and some metric usage is flawed.
Evidence
DensMathew/DensMathew/insurance_fraud_detection.ipynb: GridSearchCV over LogisticRegression, RandomForestClassifier, GradientBoostingClassifier and subsequent evaluation
DensMathew/DensMathew/Credit Card Client Classification.ipynb: create_model and train_evaluvate_model implementing a TensorFlow Sequential classifier
Business Insight & Impact
2/10
Turning analysis into business value
Some project-level conclusions and high-level next steps are written, but there is little linkage to concrete business KPIs, cost-of-error analysis, or deployment considerations for stakeholders.
Reproducibility & Notebook Hygiene
2/10
Clean, repeatable analysis
Notebooks include some reproducibility primitives (random_state, model.save) but lack pinned environment, dependency manifest, data versioning, CI, or modularized pipeline code for production.
Industries
Data & Analytics• Middle
Financial Services• Middle
Technologies
Python• Middle
Tableau
Jupyter Notebook
Power BI
Scikit-learn
Seaborn
Matplotlib
Plotly
Pandas
NumPy
Keras
TF-Keras
Recommendations
- Develop prototype classification systems and experimental model pipelines for fraud detection or credit-risk scoring including clear data contracts and unit tests.
- Focus on robust model validation and error-cost analysis (class imbalance handling, calibration, FP/FN business cost) to improve decisioning for stakeholders.
- Add reproducibility and deployment hygiene: requirements.txt or environment.yml, data versioning (DVC), modular training scripts, and small CI pipelines.
- Work on feature-leakage audits and more rigorous statistical tests to strengthen conclusions drawn from EDA and model results.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: