Computer VisionDeep LearningMedical AIImage Classification
Malaria Cell Image Classification
11/2024 - 11/2025Production-grade deep learning pipeline for automated malaria cell image classification with reproducible experiments and deployable inference services.
Description
- Developed a production-ready deep learning pipeline for malaria cell image classification
- Streamlined data management and experiment reproducibility using DVC and Git
- Deployed REST APIs for real-time inference and batch predictions
- Ensured maintainability and scalability through modular architecture and documentation
Practical Work
- Designed modular Python package architecture
- Implemented transfer learning with ResNet18
- Automated data versioning with DVC
- Built professional command-line interface
- Developed containerized Flask REST API
- Wrote unit tests and enforced code quality
Key Achievements
96.3% AccuracyReproducible PipelineProduction Ready
Technologies
PythonPyTorchFlaskDVCDockerGitGitHubVS CodePowerShell
Machine LearningMLOpsTime PredictionData Science
Flaschen Depot Delivery Time Prediction
—End-to-end machine learning solution for predicting delivery service times with automated training, evaluation, and production-ready APIs.
Description
- Designed an end-to-end ML solution for predicting delivery service times
- Built a production-ready FastAPI backend with monitoring endpoints
- Automated model training, evaluation, and deployment workflows
- Focused on reproducibility, maintainability, and real-world deployment
Practical Work
- Built modular data pipeline for loading and feature engineering
- Trained LightGBM regression model for service time prediction
- Tracked experiments and models using MLflow
- Developed FastAPI REST endpoints for predictions and metrics
- Implemented automated API tests for reliability
Key Achievements
High Model AccuracyAutomated MLOpsProduction Ready
Technologies
PythonLightGBMFastAPIMLflowPandasNumPyPytestJupyter NotebookGit
Data EngineeringETLData AnalyticsVisualization
UNICEF ETL & Analytics Pipeline
—Scalable ETL and analytics pipeline for processing, analyzing, and visualizing donation and campaign data for actionable insights.
Description
- Developed ETL and analytics pipeline as part of a UNICEF technical challenge
- Processed and analyzed donation and campaign datasets
- Focused on clarity, reproducibility, and scalability
- Enabled data-driven insights for fundraising activities
Practical Work
- Implemented data ingestion and preprocessing scripts
- Built robust data cleaning and transformation routines
- Performed exploratory data analysis on donation data
- Created clear visualizations for insights communication
- Compiled reproducible analysis reports in Jupyter Notebooks
Key Achievements
Automated ETLActionable InsightsProfessional Reporting
Technologies
PythonPandasNumPySQLJupyter NotebookMatplotlibSeabornGitVS Code
Predictive MaintenancePredictiveMaintenance
Predictive Maintenance for Industrial Equipment
04/2022 - 06/2022Custom LLM fine-tuned on legal domain with RAG-enhanced document processing pipeline.
Description
- Assist in predictive analytics tools screening
- Design and implement predictive models with R and SAS
- Provide predictive analytics expertise
- Support platform development and deployment
Practical Work
- Built real-time invoice assessment system
- Optimized API data processing
- Implemented ML algorithms for fraud detection
Key Achievements
Predictive modelingWeight optimizationVolume forecasting
Technologies
ReactTypeScriptTailwind CSSReactyTypeScriptyTailwind CSSy