AI Engineer Internship (Computer Vision)

• Build vision solutions: write Python and OpenCV code, and work with YOLO / Ultralytics and other deep learning models for detection, classification, tracking, recognition, OCR and video analytics.
• Handle the data: prepare, clean, organize and annotate image and video datasets, then train, fine-tune, validate and evaluate models.
• Work with live video: build image and video pipelines, including CCTV footage, RTSP streams and real-time processing.
• Measure and improve models: use precision, recall, mAP, F1 score and IoU, study false positives and false negatives, and check real-world performance instead of trusting training metrics alone.
• Integrate and deploy: put trained models into Python applications and APIs (Flask or FastAPI where needed), debug inference pipelines, and help deploy them. Docker, Linux, cloud platforms, CUDA, ONNX or TensorRT may come up when applicable.
• Research and document: read papers, documentation and GitHub repositories, build small proofs of concept, explain why one approach was chosen, and record experiments and decisions.

The post does not list a degree, year of study or experience level. What it makes clear is the skill set: Python, OpenCV and image recognition are the skills named for this role.

It also describes a way of working. You are expected to research problems on your own before asking for help, read technical material to find solutions, and troubleshoot model and pipeline issues step by step. The work mode is remote.

• Keep a small project that uses Python and OpenCV, ideally with a YOLO model trained or fine-tuned on a dataset you annotated yourself.
• Show evaluation, not just results. It helps to report precision, recall, mAP and IoU, and to write a short note on why your model produced false positives.
• Try reading frames from a video file or a stream and running detection on them, since this role involves real-time video.
• Wrap a model in a simple Flask or FastAPI endpoint and keep the code on GitHub with a clear README.
• Practise explaining a choice you made between two approaches and what you tested to decide.