Your software engineering portfolio looks like everyone else's.

Build one 3D project that stands out. The free 3D Portfolio Project Starter for early-career engineers aiming at robotics, AR/VR, games, mapping and CAD teams.

The free 3D Portfolio Project Starter

A diffusion model that generates 3D shapes: it starts from random points and removes the noise step by step until a chair or an airplane appears. You build the parts an interviewer will ask about:

• Data pipeline: load 3D meshes, sample them into point clouds, normalize and visualize them.

• Nearest-neighbor search: brute force first, then timed against a KD-tree.

• Diffusion loop: the noising step and the sampling loop that turns noise into a shape.

• Evaluation: Chamfer distance on your own nearest-neighbor code, plus a best/worst gallery.

Python, NumPy, PyTorch, SciPy KD-trees, Open3D, public 3D datasets. Trains on a free Colab GPU.

Why this project

• 3D and generative AI, not CRUD. The exact vocabulary a spatial AI interviewer probes.

• Explains in one sentence: it learns to remove noise step by step, so random points become a chair.

• Yours to extend: class conditioning, faster sampling, or shape completion from a partial LiDAR-style scan.

Who this is for

CS and engineering students, new grads, and career switchers aiming at robotics, AR/VR, games, autonomy, mapping and CAD. If you want AI news roundups, this is the wrong list.

I am a senior ML engineer working on 3D and spatial AI, and I have taught early-career engineers for over ten years.