Project Ideas
Hands-on projects spanning neural simulation, synthetic data, sensor modeling, and agent behavior. Organized into three tracks aligned with Applied Intuition's core technology areas.
Neural Simulation
Reconstruct and render driving scenes using neural representations — 3D Gaussian Splatting, NeRF, and differentiable rendering.
3DGS Scene Reconstruction
Multi-Sensor Neural Renderer
Dynamic Scene Decomposition
Neural Sim Quality Evaluator
Synthetic Data & Sensor Sim
Generate training data at scale and simulate sensor physics — camera, lidar, and radar with physically accurate models.
Synthetic Data Pipeline
Domain Adaptation Benchmark
Physics-Based Sensor Simulator
Minority Class Augmentation
Agent Behavior & Infrastructure
Build realistic sim agents, scale simulation infrastructure, and evaluate AD systems with closed-loop testing.
Sim Agent with Realism Metrics
Distributed Simulation Pipeline
Adversarial Scenario Generator
Closed-Loop Evaluation System
Tips for Success
Make the most of these projects — build skills aligned with Applied Intuition's simulation technology stack
Start with Metrics
Begin with the Neural Sim Quality Evaluator to understand how simulation quality is measured before building renderers.
Benchmark Rigorously
Include quantitative comparisons — PSNR, SSIM, FPS, throughput — to demonstrate engineering rigor.
Cross Tracks
Combine projects across tracks — e.g., use neural rendering output as input for synthetic data pipelines.
Read the Deep Dives
Each project maps to one or more deep dive papers — read the neural rendering and sensor sim papers first.
Ready to Build?
Start with the deep dive papers to build foundational knowledge, then pick a project track that aligns with your interests.