ML / Computer Vision Intern

Full Job Description

What You Will Work On

Annotate image datasets using Label Studio — bounding boxes across multiple object classes; maintain annotation consistency and quality.
Train and tune YOLOv8 object detection models; iterate on hyperparameters, anchor settings, and input resolution to hit target accuracy benchmarks.
Evaluate model performance using standard metrics — mAP50, precision, recall, confusion matrix — and produce structured analysis reports.
Export trained models to ONNX format and benchmark inference performance for CPU-only deployment.
Design and apply data augmentation strategies (geometric transforms, noise injection, compression simulation) to improve generalisation on limited training data.
Integrate PaddleOCR to extract text from regions of interest identified by the detection model.
Implement spatial association logic to link detected objects with nearby text using bounding box geometry and proximity heuristics.
Build post-processing routines that clean, normalise, and validate raw OCR output before it enters the structured data pipeline.
Run end-to-end accuracy evaluations; identify and categorise error sources across detection, OCR, and association stages.

Must-Have Skills

Python 3.10+PyTorchUltralytics YOLOOpenCVObject detection conceptsmAP / Precision / RecallString parsing & regexGit + command line

Nice to Have

PaddleOCR or TesseractLabel StudioONNX RuntimeAlbumentationsNumPy / matplotlibPyMuPDF / fitzLinux / terminal workflow

Shared Expectations — All Roles

Show up and contribute daily. This is an in-office role. You are expected to be present, engaged, and making progress you can show — not just busy.
Weekly deliverables. Every week you should have something reviewable — code, annotated data, evaluation results, or a written finding.
Own your errors. If something you built breaks downstream, you debug it and you write the test that catches it next time.
Understand what you build. Develop familiarity with the industry domain — why accuracy of the output matters, not just that the code runs.
Use AI tools responsibly. Using AI to assist is fine. Submitting output you cannot explain in a code review is not.

Apply Online

Submit Your Application

First Name

Last Name

Gender

Male

Female

Mobile

Highest Qualification

Date of Birth

Expected Stipend (optional)

Applying For

Resume / CV PDF, DOC, DOCX — max 2 MB

QONG Systems evaluates candidates solely on demonstrated skill and potential.
No degree requirement — final-year students, bootcamp graduates, and self-taught engineers are all welcome.

ML / Computer Vision Intern

Shared Expectations — All Roles

Ready to Apply?

Submit Your Application