I work on machine learning that can be understood and tested.
I studied computer engineering at Purbanchal University and became interested in machine learning through projects that bring together software, data and practical problems.
What I work on
Graph learning and fraud detection. I represent transactions and accounts as connected networks, then use graph structure and node embeddings to look for behaviour that may be missed when records are analysed one at a time. Tools include NetworkX, Neo4j and Node2Vec.
Explainable prediction. I use SHAP to examine which features contribute to predictions in health-related and financial datasets. An explanation does not prove that a model is correct, but it makes the model behaviour easier to review.
Computer vision. I have worked on prototypes for drowsiness detection, object tracking and medical image classification using OpenCV, YOLOv8 and PyTorch.
Alongside model development, I am building my software engineering skills through project structure, APIs, testing, containers and model monitoring.
My toolbox
Languages
- Python
- JavaScript
- C++
- C
- SQL
- Bash
Machine learning
- scikit-learn
- PyTorch
- PyTorch Geometric
- TensorFlow
- OpenCV
- YOLOv8
- pandas
- NetworkX
- Node2Vec
- SHAP
Application and development tools
- FastAPI
- Django
- Streamlit
- React
- Node.js
- Neo4j
- Git
Currently learning and developing
- Docker
- MLflow
- pytest
- CI/CD
- Model monitoring
How I work
Keep work reproducible
I document setup steps, dependencies and assumptions so that the work can be reviewed and repeated.
Choose useful metrics
I select evaluation metrics based on the problem, look beyond accuracy when classes are imbalanced and record the limitations.
Be clear about limitations
A prototype is not the same as a deployed service, and a model estimate is not a guarantee. I try to make the status and limitations clear.
Awards and recognition
Best System Award
For the Graph-Based Fraud and Anomaly Detection System.
Best Project Proposal (2024)
For the heart disease prediction system.
Leadership
AI/ML learning and experimentation
Led a learning and experimentation series and developed small AI prototypes for workflow automation and interactive interfaces.
I am interested in roles involving applied machine learning, data-driven software and ML engineering. I would like to contribute to practical projects while continuing to strengthen my engineering skills.