Nassim B.

🔹 Customer Ticket Pipeline & Churn Prediction

Created automated pipelines with Airflow, DBT, and BigQuery to process and structure tens of thousands of customer tickets.

Detected weak churn signals and triggered proactive actions through SQL analysis and simple Python models.

🔹 Feedback Analysis & Product Irritant Reduction

Ingested and modeled customer feedback (tickets, NPS) to identify friction points using DBT, SQL, and BigQuery.

Delivered actionable recommendations via Looker dashboards and automated reports with Airflow.

🔹 Real-time Monitoring Dashboards (Support & Product)

Built interactive dashboards (KPIs, SLAs) with Looker, Data Studio, and Power BI, updated in near real-time.

Orchestrated data collection and transformations using Python, SQL, and tools like Airflow / in-house ETL.

As part of this project, I designed a computer vision system to automatically detect cars in complex scenes (crowded parking lots, urban traffic, difficult weather conditions, etc.). The objective was to create a robust solution capable of recognizing and locating vehicles in heterogeneous images, considering partial occlusion, variable scale, and disturbed visual environments.

Python, PyTorch, OpenCV

YOLOv5, Faster R-CNN, SSD

LabelImg for annotation, TensorBoard for training monitoring

✅ Result:

Accurate vehicle detection with an mAP of 95% on complex scenes.

Reusable pipeline for other object detection use cases (pedestrians, traffic lights, signs, etc.).

Demonstrable implementation in a notebook or demo API.