Machine learning & data science

Build image models (detection/classification)

OpenCV + pretrained CNN/YOLO fine-tuning for a practical vision task.

~20 focused hours·intermediate

Tools: OpenCV, PyTorch/torchvision, YOLO (Ultralytics), Roboflow/LabelImg

Market relevance — share of job ads asking for this
What employers mean

You should be able to…

  1. Preprocess images with OpenCV (resize, normalize, augment, color space conversions)
  2. Fine-tune a pretrained CNN (ResNet/EfficientNet) for an image classification task
  3. Fine-tune or run inference with a detection model (YOLO) for object detection
  4. Prepare and validate an annotated image dataset (bounding boxes) in the right format
  5. Evaluate detection/classification with the right metrics (mAP, IoU, precision/recall)
  6. Handle real-world image quality issues (lighting, blur, occlusion, low resolution)
  7. Optimize a vision model for reasonable inference latency, not just accuracy

Needs first: Build and train neural networks in PyTorch

Learn — free, link-checked

The few resources that matter

Read · intermediate · 60 min · docs.ultralytics.com

YOLO Object Detection & Segmentation

The docs for the most-used open detection model in Indian CV job postings -- train and evaluate a custom detector with a few lines of code. — Ultralytics
Read · intermediate · 150 min · docs.opencv.org

OpenCV-Python Tutorials

Official image-processing fundamentals (thresholding, contours, feature detection) that every CV pipeline needs before a model even runs. — OpenCV
Course · intermediate · 180 min · kaggle.com

Computer Vision

Notebook-based course on CNNs, transfer learning and data augmentation with an immediate Kaggle submission to prove the result. — Kaggle Learn
Course · intermediate · 1200 min · course.fast.ai

Practical Deep Learning for Coders

Top-down, code-first course (free videos + notebooks) that gets you to a fine-tuned, deployed image classifier before it explains the underlying math. — fast.ai
Practice

Indian vehicle/license-plate style object detector

Build an object detector (fine-tune YOLO via Ultralytics) to detect and localize a practical Indian-context object class, such as vehicle types (auto-rickshaw, two-wheeler, car) or license plates, using a small annotated dataset you label yourself with Roboflow/LabelImg or a public dataset. Evaluate with mAP and test on a handful of real photos you take yourself.

Done when
  • Annotated dataset (100+ labeled images minimum) in YOLO format with a documented class list
  • YOLO model fine-tuned and evaluated with mAP@0.5 reported on a held-out validation split
  • Inference demonstrated on at least 5 real-world test images not in the training set, with visualized bounding boxes
  • README documents failure cases observed (e.g. poor lighting, occlusion) and what would improve them
Prove it

Evidence a recruiter can check

  • Public GitHub repo with dataset annotation format, training config and mAP results
  • Before/after example images with predicted bounding boxes checked into the README
  • Exported model weights (or Hugging Face Hub upload) others can run inference with
  • Can explain live the mAP metric and a specific failure case observed
Interview

Questions you'll get asked

  1. Walk me through fine-tuning a pretrained model for a new image classification task
  2. How is object detection different from classification, and what does mAP measure?
  3. How would you build a system to detect defective products on a manufacturing line from images?
  4. What data augmentation would you use for a small image dataset, and why?
  5. How do you handle class imbalance in an object detection dataset (rare object classes)?
  6. What's IoU, and how does it factor into evaluating a detection model?
  7. How would you reduce inference latency for a vision model running on limited hardware?
See where you stand for Machine Learning Engineer