← Back to Work

Continuous Intelligence: Neural ODEs for Modelling Robot Dynamics

Built a Neural ODE to predict trajectories of a 7-DOF Baxter arm from given input torques; used PyTorch and wrote custom integrators; benchmarked integrators using RMSE visualizations.

Timeframe
Sept – Dec 2025
Role
Data Analysis, Machine Learning, Dynamical Modelling
Org / Context
Course Project

Problem

[What the team needed and why — keep this at a level your employer would be comfortable with public; describe the problem, not proprietary specifics.]

Approach

[Your actual engineering decisions and trade-offs, at a level appropriate to share publicly. Save proprietary detail for the interview conversation itself.]

My Contribution

[What you specifically built or owned on the team — be explicit even though it was a team/company effort.]

Result

[Quantify wherever possible and appropriate to share: cycle time, accuracy, cost, latency, adoption.]

Stack

  • PyTorch
  • torchdiffeq

Links

What I'd Do Differently

[Optional but strong — shows self-awareness and engineering maturity.]