AVAILABLE FOR REMOTE WORK

EMMANUEL
JOHN LELO

Full-Stack Data Scientist  ·  MLOps Engineer

🌍 Open to Remote — Europe & Worldwide

Building intelligent, production-ready ML systems — from exploratory analysis and model training to CI/CD pipelines, containerised APIs, and Kubernetes-scaled deployments.

🚀 View Projects ✉️ Get in Touch
6+
Projects
99.9%
Best Accuracy
5+
Certifications
2
MLOps Systems
Who I Am

About Me

Full-Stack Data Scientist and MLOps Engineer passionate about shipping production ML systems.

🧑‍💻
Emmanuel John Lelo
Data Scientist · MLOps Engineer
Python MLOps Kubernetes FastAPI Docker
📍 LocationDar-es-Salaam, Tanzania
📧 Emailleloemmanuel540@gmail.com
💼 Experience1 Year
🌍 Work ModeRemote · Europe & Worldwide

Hello, I'm Emmanuel 👋

I'm a Full-Stack Data Scientist and MLOps Engineer with 1 year of hands-on experience building real ML systems — not just notebooks. I specialise in the full ML lifecycle: from raw data and feature engineering through to containerised model serving, CI/CD automation, and production monitoring.

I've shipped two production MLOps pipelines (bank churn + credit card fraud), completed virtual experience programmes with British Airways and Lloyds Bank Group, and built a deep learning fraud detection system achieving 99.9% accuracy on 284k transactions.

My toolkit spans Python, Scikit-learn, TensorFlow, XGBoost, DVC, MLflow, Prefect, Docker, Kubernetes, FastAPI, AWS, and more. I'm passionate about turning data into reliable, scalable production systems.

🌍

Open to Remote Opportunities — Europe & Worldwide

Actively seeking remote roles as a Data Scientist or MLOps Engineer. Available for full-time, part-time, or contract engagements. Comfortable with async workflows and international time zones.

What I Work With

Technical Skills

A full-stack toolkit spanning data science, ML engineering, and production infrastructure.

🐍
Programming & Data
PythonSQLRPandasNumPy
🧠
Machine Learning & AI
Scikit-learnTensorFlowXGBoostLightGBMHuggingFace
⚙️
MLOps & DevOps
MLflowDVCPrefectGitHub ActionsZenML
☁️
Cloud & Infrastructure
AWS SageMakerAWS S3DockerKubernetesHelm
📊
Analysis & Visualisation
PostgreSQLMongoDBPlotly / DashPower BIStreamlit
🔍
Model Serving & Monitoring
FastAPIPrometheusGrafanaEvidently AIPytest
Skill Depth

My Proficiency

Honest self-assessment of core technical competencies.

Python & Data Science Libraries88%
SQL & Data Analysis85%
Machine Learning & Model Development82%
MLOps Pipelines & CI/CD75%
Docker & Kubernetes72%
AWS Cloud Platform70%
Model Monitoring & Observability70%
What I've Built

Work Experience

A track record of shipping real ML systems, not just prototypes.

🔐
MLOps Engineer — Production Churn Prediction System
2024 · Self-Initiated
Self-Initiated · bank-churn-mlops · github.com/emmanuelmassawe200
  • Architected a production-grade MLOps pipeline covering DVC + DagsHub data versioning, MLflow experiment tracking, Prefect orchestration, and a containerised FastAPI REST API serving real-time predictions.
  • Published Docker image (emmanuely2000/bank-churn-api), achieving 84% accuracy and 63.1% weighted F1 with a scikit-learn Pipeline bundling preprocessing and Random Forest as a single serialisable artifact.
  • Implemented full CI/CD via GitHub Actions — automated test → build → push to Docker Hub on every commit, reducing manual deployment steps to zero.
  • Resolved production challenges: scikit-learn version mismatch across training/serving, Docker secrets via runtime env vars, and FastAPI CORS for browser clients.
  • Roadmapped Evidently drift detection, Prefect scheduled retraining, SHAP explainability, GCP Cloud Run deployment as next-phase additions.
💳
MLOps Engineer — Real-Time Credit Card Fraud Detection
2024 · Self-Initiated
Self-Initiated · credit-card-fraud-detection · github.com/emmanuelmassawe
  • Built a production fraud detection system on 284,807 transactions (0.17% fraud rate), achieving 99.9% accuracy, 93.2% F1, 98.7% ROC-AUC with XGBoost + SMOTE; FastAPI serving with sub-100ms response time.
  • Deployed on Kubernetes (Minikube + GKE/EKS/AKS-ready) with HPA autoscaling, rolling updates, and health-check probes — scaling from 1 to 5+ replicas under load.
  • Implemented complete MLOps lifecycle: DVC versioning, MLflow + DagsHub tracking, Prefect orchestration, dual GitHub Actions workflows with 90%+ test coverage via Pytest.
  • Engineered a modular pipeline across 5 reproducible DVC stages (ingestion → preprocessing → training → evaluation → serving) with full dvc repro traceability.
  • Delivered browser UI with real-time confidence scores; roadmapped Prometheus/Grafana monitoring and SHAP explainability.
✈️
Data Scientist — Lounge Demand & Revenue Modelling
2024 · Virtual Experience
British Airways · Virtual Experience (Forage)
  • Analysed a 50,000+ flight schedule dataset to model lounge eligibility across 3 tiers — identified peak demand at ~1,280 users/hour at 07:00–08:00.
  • Built a revenue & cost model projecting £262.8M annual profit; recommended Concorde Room expansion with projected 2-month payback on £7–10M investment.
  • Engineered a 6-panel Matplotlib/Seaborn dashboard + multi-sheet Excel workbook; model predictions within 3% of actual BA data (long-haul 41% vs actual 43.9%).
  • Delivered a one-page executive summary with dynamic staffing recommendation saving ~£500k/year.
📈
ML Engineer — Customer Booking Conversion Prediction
2024 · Virtual Experience
British Airways · Virtual Experience (Forage)
  • Built a binary classification pipeline on 50,000 customer records predicting booking completion; achieved recall of 0.62 on a heavily imbalanced dataset (15% positive class).
  • Engineered 8 domain-specific features (booking lead, group size, long-haul flag, extras score); applied SMOTE + tuned Random Forest & XGBoost via GridSearchCV/RandomizedSearchCV.
  • Identified booking origin (34% importance) and trip duration (20%) as top predictors; recommended strategy projected to deliver 5–10% conversion lift.
🏦
Data Scientist — Bank Customer Churn Prediction
2024 · Self-Initiated
Lloyds Bank Group · Self-Initiated Project
  • Compared 4 classifiers (Logistic Regression, Random Forest, XGBoost, SVM) on 10,000+ customer records; engineered 4 behavioural features and applied SMOTE + 5-fold StratifiedKFold CV.
  • Tuned Random Forest (72 combinations) and XGBoost (24 combinations) via GridSearchCV; findings applicable to retention campaign targeting for high-risk segments.
🧠
ML Engineer — Financial Fraud Detection (Deep Learning)
2024 · Self-Initiated
Self-Initiated · PaySim Synthetic Dataset
  • Designed a TensorFlow/Keras ANN (128→64→1, dual Dropout 0.5) on large-scale synthetic payments data; implemented precision-recall threshold optimisation over default 0.5 cutoff.
  • Scaffolded sklearn Pipeline architectures for Logistic Regression, XGBoost, LightGBM with integrated SMOTE and RandomizedSearchCV for multi-model benchmarking.
  • Implemented EarlyStopping callbacks, SMOTE + MinMaxScaler, and joblib serialisation for deployment-ready inference.
Portfolio

Featured Projects

Production-ready ML systems with measurable impact and modern engineering practices.

🔐
MLOPS · KUBERNETES · FASTAPI
Credit Card Fraud Detection Pipeline
99.9% ACC · 93.2% F1 · 98.7% AUC
Production-grade fraud detection on 284k transactions — XGBoost + SMOTE, Kubernetes HPA autoscaling, dual GitHub Actions CI/CD, 90%+ test coverage via Pytest.
XGBoostKubernetesFastAPIDVCMLflowPrefect
🏭
MLOPS · DOCKER · CICD
Bank Churn MLOps Pipeline
84% ACC · 63.1% F1 · Zero-touch deploy
End-to-end MLOps system: DVC data versioning, Prefect orchestration, MLflow registry, FastAPI serving, Docker Hub distribution, GitHub Actions CI/CD.
Random ForestDockerMLflowDVCDagsHubFastAPI
✈️
DATA SCIENCE · REVENUE MODELLING
British Airways Lounge Analysis
£262.8M projected profit · 3% model error
Demand forecasting and revenue modelling for BA Terminal 3 lounge operations — 50k+ records, 6-panel dashboard, stakeholder executive summary.
PandasMatplotlibSeabornExcelNumPy
📈
CLASSIFICATION · FEATURE ENGINEERING
BA Booking Conversion Prediction
0.62 Recall · 8 engineered features
Binary classification on 50k records to predict flight booking completion — SMOTE, 8 engineered features, Random Forest + XGBoost, stakeholder presentation.
XGBoostRandom ForestSMOTEscikit-learn
🏦
CLASSIFICATION · CHURN ANALYSIS
Lloyds Bank Churn Prediction
4 models compared · 72 param combos tuned
End-to-end churn prediction pipeline comparing LR, Random Forest, XGBoost, and SVM — 4 engineered features, SMOTE, 5-fold StratifiedKFold CV.
XGBoostSVMSMOTEGridSearchCV
🧠
DEEP LEARNING · ANN · FRAUD
Synthetic Financial Fraud Detection
ANN 128→64→1 · Threshold-tuned F1
TensorFlow/Keras ANN with Dropout regularisation, precision-recall threshold optimisation, SMOTE + MinMaxScaler pipeline, and joblib deployment-ready serialisation.
TensorFlowKerasXGBoostLightGBMSMOTE
Credentials

Certifications & Education

Continuous learning backed by industry-recognised credentials.

☁️
AWS Fundamentals
Amazon Web Services · 2024
⚙️
MLOps Specialization
DeepLearning.AI / Coursera · 2025
🧠
Deep Learning Specialization
DeepLearning.AI / Coursera · 2024
🐳
Docker & Kubernetes Fundamentals
Linux Foundation · 2024
📊
Data Science Professional Certificate
IBM / Coursera · 2023
🎓
B.Sc. Data Science
EASTC University · Tanzania
Let's Connect

Get In Touch

Open to remote roles worldwide. I reply within 24 hours.

I'm actively seeking remote opportunities in Europe and worldwide as a Full-Stack Data Scientist or MLOps Engineer. Whether you have a role, project, or just want to connect — drop me a message.

📧
leloemmanuel540@gmail.com
💼
LinkedIn Profile
🐙
github.com/emmanuelmassawe
🔬
DagsHub — emmanuelmassawe200

✉️ Send Me a Message

Emmanuel John Lelo
Full-Stack Data Scientist & MLOps Engineer
📧 leloemmanuel540@gmail.com
📱 +255 696 393 526
💼 linkedin.com/in/emmanuel-john-61b343334
🐙 github.com/emmanuelmassawe
📍 Dar-es-Salaam, Tanzania
🌍 Open to Remote — Europe & Worldwide 1 Year Experience Python · ML · MLOps · Kubernetes
⚡ Core Skills
Python & Data Libraries88%
Machine Learning82%
SQL & Data Analysis85%
MLOps & CI/CD75%
Docker & Kubernetes72%
AWS Cloud70%
🛠️ Tools & Frameworks
Scikit-learnTensorFlowXGBoost LightGBMHuggingFaceMLflow DVCDagsHubGitHub Actions PrefectFastAPIEvidently AI PrometheusGrafanaDocker KubernetesAWS SageMakerPytest
🎓 Education
B.Sc. Data Science
EASTC · Dar-es-Salaam
📜 Certifications
▸ AWS Fundamentals
▸ MLOps Specialization — DeepLearning.AI
▸ Deep Learning Specialization — DeepLearning.AI
▸ Docker & Kubernetes — Linux Foundation
▸ Data Science Certificate — IBM
🗣️ Languages
▸ English — Fluent
▸ Swahili — Native
👤 Professional Summary

Results-driven Full-Stack Data Scientist & MLOps Engineer with 1 year of hands-on experience across data analysis, ML model development, and production deployment pipelines. Shipped 2 end-to-end MLOps systems (bank churn + credit card fraud) achieving up to 99.9% accuracy. Skilled in DVC, MLflow, Docker, Kubernetes, FastAPI, and GitHub Actions CI/CD. Actively seeking remote opportunities in Europe and worldwide.

💼 Experience & Projects
MLOps Engineer — Credit Card Fraud Detection
2024
Self-Initiated · Kubernetes · XGBoost · FastAPI
  • 284,807 transactions; 99.9% accuracy, 93.2% F1, 98.7% ROC-AUC. Sub-100ms FastAPI inference.
  • Full Kubernetes deployment (HPA autoscaling, rolling updates) + dual GitHub Actions CI/CD, 90%+ test coverage.
  • Complete MLOps: DVC · MLflow · DagsHub · Prefect · 5-stage reproducible pipeline.
MLOps Engineer — Bank Churn Prediction System
2024
Self-Initiated · Docker · MLflow · Prefect
  • 84% accuracy; Docker image published to Docker Hub; zero-touch CI/CD via GitHub Actions.
  • DVC + DagsHub data versioning; Prefect orchestration; FastAPI REST serving.
Data Scientist — British Airways (Forage)
2024
Lounge Demand Modelling + Booking Conversion Prediction
  • £262.8M profit model from 50k+ records; predictions within 3% of BA actual data.
  • Booking conversion: 0.62 recall on 50k records; 8 engineered features; Random Forest + XGBoost.
Data Scientist — Lloyds Bank & Financial Fraud (Deep Learning)
2024
Churn Prediction · ANN Fraud Detection · scikit-learn · TensorFlow
  • 4 classifiers compared (LR, RF, XGBoost, SVM); 72+24 hyperparameter combinations tuned via GridSearchCV.
  • TensorFlow ANN (128→64→1, Dropout 0.5); precision-recall threshold optimisation; joblib serialisation.
🚀 Key Projects
🔐 Fraud Detection (K8s)
99.9% acc · Kubernetes HPA · sub-100ms API
XGBoostKubernetesFastAPI
🏭 Bank Churn MLOps
84% acc · Docker · Zero-touch CI/CD
MLflowDVCPrefect
✈️ BA Lounge Analysis
£262.8M model · 3% prediction error
PandasSeaborn
🧠 ANN Fraud Detection
TF/Keras · Threshold-tuned F1 · joblib
TensorFlowLightGBM
Emmanuel John Lelo
Full-Stack Data Scientist & MLOps Engineer
🌍 Open to Remote — Europe & Worldwide 1 Year Experience Python · ML · MLOps · Kubernetes
📧 leloemmanuel540@gmail.com
📱 +255 696 393 526
📍 Dar-es-Salaam, Tanzania
💼 linkedin.com/in/emmanuel-john-61b343334
🐙 github.com/emmanuelmassawe
🔬 dagshub.com/emmanuelmassawe200
👤 Professional Profile
I'm a dedicated Full-Stack Data Scientist and MLOps Engineer with 1 year of professional experience spanning data analysis, machine learning, and production ML systems. I've shipped 2 end-to-end MLOps pipelines achieving up to 99.9% accuracy, deployed on Kubernetes with full CI/CD, experiment tracking, and data versioning. I cover the complete ML lifecycle — EDA, feature engineering, model training, containerised serving, and monitoring. Actively seeking remote opportunities in Europe and worldwide.
💼 Experience & Projects
MLOps Engineer — Credit Card Fraud Detection System
2024 · Self-Initiated
github.com/emmanuelmassawe/credit-card-fraud-detection
  • Production fraud detection on 284,807 transactions: 99.9% accuracy, 93.2% F1, 98.7% ROC-AUC with XGBoost + SMOTE. Sub-100ms FastAPI inference.
  • Kubernetes deployment (Minikube + GKE/EKS/AKS) with HPA autoscaling, rolling updates, health probes.
  • Full MLOps: DVC versioning, MLflow + DagsHub tracking, Prefect orchestration, dual GitHub Actions workflows, 90%+ Pytest coverage.
  • 5-stage reproducible DVC pipeline: ingestion → preprocessing → training → evaluation → serving.
MLOps Engineer — Bank Churn Prediction Pipeline
2024 · Self-Initiated
emmanuely2000/bank-churn-api · Docker Hub
  • Production MLOps pipeline: DVC + DagsHub data versioning, MLflow experiment tracking, Prefect orchestration, FastAPI serving.
  • 84% accuracy, 63.1% weighted F1. Docker image published; GitHub Actions CI/CD (zero manual deployment).
  • Resolved production challenges: sklearn version mismatch, Docker secrets management, FastAPI CORS.
Data Scientist — British Airways Virtual Experience (Forage)
2024
  • Lounge Analysis: Modelled demand on 50k+ records; £262.8M revenue model, predictions within 3% of BA actual data. 6-panel Seaborn dashboard + Excel workbook.
  • Booking Prediction: 0.62 recall on 50k records; 8 engineered features; SMOTE + RF/XGBoost; booking origin = 34% feature importance.
Data Scientist — Lloyds Bank Churn + Financial Fraud (Deep Learning)
2024
  • Lloyds Churn: 4 classifiers on 10k+ records; 4 engineered features; 72+24 GridSearchCV combinations; StratifiedKFold CV.
  • ANN Fraud Detection: TF/Keras ANN (128→64→1, Dropout 0.5) on PaySim dataset; precision-recall threshold optimisation; sklearn Pipelines for LR, XGBoost, LightGBM.
⚡ Technical Skills
🐍 Programming & Data
PythonSQLRPandasNumPy
🧠 ML & Deep Learning
Scikit-learnTensorFlowXGBoostLightGBMHuggingFace
⚙️ MLOps & DevOps
MLflowDVCDagsHubPrefectGitHub Actions
☁️ Cloud & Infrastructure
AWS SageMakerAWS S3DockerKubernetesHelm
📊 Analysis & Visualisation
PostgreSQLMongoDBPlotlyPower BIStreamlit
🔍 Serving & Monitoring
FastAPIEvidently AIPrometheusGrafanaPytest
🚀 Featured Projects
🔐
Fraud Detection (Kubernetes)
99.9% acc · 93.2% F1 · HPA autoscaling · 90%+ test coverage · sub-100ms API
XGBoostKubernetesFastAPIMLflow
🏭
Bank Churn MLOps Pipeline
84% acc · Docker Hub · Zero-touch CI/CD · DVC + MLflow + Prefect
MLflowDVCDockerPrefect
✈️
BA Lounge Revenue Model
£262.8M projection · 50k+ records · 3% model error · 6-panel dashboard
PandasSeabornMatplotlib
📈
BA Booking Conversion
0.62 recall · 8 features engineered · 50k records · SMOTE
XGBoostSMOTEscikit-learn
🏦
Lloyds Bank Churn
4 models · 72 GridSearch combos · StratifiedKFold · SMOTE
XGBoostSVMGridSearchCV
🧠
ANN Fraud Detection
TF/Keras 128→64→1 · Dropout · Threshold-tuned · joblib deploy
TensorFlowLightGBMSMOTE
🎓 Education
🎓
B.Sc. Data Science
EASTC University · Tanzania
ML · Algorithms · Database Systems · Statistics
📚
Self-Directed Learning
Coursera · DeepLearning.AI · Kaggle · AWS
2023 – 2025
📜 Certifications
☁️
AWS Fundamentals
Amazon Web Services · 2024
⚙️
MLOps Specialization
DeepLearning.AI · 2025
🧠
Deep Learning Specialization
DeepLearning.AI · 2024
🐳
Docker & Kubernetes
Linux Foundation · 2024
📊
Data Science Certificate
IBM / Coursera · 2023
🏅
Machine Learning with Python
freeCodeCamp · 2023
🏆 Key Achievements
99.9%
Accuracy on credit card fraud detection with 284,807 transactions; 93.2% F1 on 0.17% imbalanced class.
2
End-to-end MLOps systems shipped and containerised, with full CI/CD, experiment tracking, and Kubernetes deployment.
90%+
Automated test coverage via Pytest on fraud detection system; dual GitHub Actions CI/CD workflows.
£262.8M
Revenue projection model for British Airways lounge operations, within 3% of actual BA data.
🗣️ Languages & Soft Skills
Languages
English
Swahili
Soft Skills
Analytical ThinkingProblem SolvingTeam CollaborationFast LearnerRemote-ReadyClear Communication
🌍 Availability
🌍

Open to Remote Opportunities — Europe & Worldwide

Seeking full-time, part-time, or contract remote roles as a Data Scientist or MLOps Engineer. Available immediately. Comfortable with async workflows and international time zones.

Emmanuel John Lelo · Full-Stack Data Scientist & MLOps Engineer · leloemmanuel540@gmail.com · 🌍 Open to Remote · 2026