PINN Training — Physics‑Informed Neural Networks for Engineering
A practical course for engineers and data scientists to build, train, and deploy PINNs for real‑world problems.
Cohort‑based (live) and on‑demand tracks. Limited seats.
Who is this for?
- • CAE / R&D engineers integrating AI with physics‑based workflows
- • Data scientists building models under PDE / constraint regimes
- • Technical leaders exploring production‑grade physics‑AI pipelines
You’ll be able to…
- • Frame PDEs for PINNs: governing equations, BC/ICs, and non‑dimensionalization
- • Diagnose & fix real‑world failure modes (stiffness, gradient pathologies, loss imbalance)
- • Choose sampling curricula & loss weighting that actually converge on tough PDEs
- • Validate against trusted solvers (FEM/CFD) and design robust error checks
- • Move from prototype to production with reproducible training and monitoring
Syllabus
Program Overview
- • Mode: Live online sessions with interactive labs
- • Dates:February 23–27, 2026
- • Duration: 5 days (with post-course lab access)
- • Certification: Certificate of Completion from oorja
- • Post-course: 3-month Cloud PINN Lab license included + optional monthly mentorship sessions
What you’ll learn
- • Build and train PINNs for solving physical systems
- • Understand loss formulation, sampling, and optimization
- • Move from 1D conceptual problems to 2D/3D domains
- • Diagnose and improve training stability and convergence
- • Apply PINNs to your own research or engineering problem
- • Access the oorja Cloud PINN Lab for 3 months to experiment, test, and deploy your models
Day 1 — Foundations of Physics-Informed Learning
Gain intuition about how neural networks learn physical laws.
Exercises: 1D heat equation, residual visualization, and adding sinusoidal sources.
Day 2 — Optimization and Training Mechanics
Learn how loss weighting, optimizers, and sampling affect convergence.
Exercises: Compare ADAM vs L-BFGS, test sampling density, and detect pathological solutions.
Day 3 — Extending to 2D/3D
Define and visualize multidimensional PDE setups.
Exercises: Simulate heat flow in 2D plates and 3D cylinders; add transient conditions.
Day 4 — Advanced Behavior and Stability
Explore how architecture and scheduling influence performance.
Exercises: Scheduler tuning, network-depth comparison, coupled-physics examples.
Day 5 — Research Applications & Mini-Project
Solve a custom problem and present your results.
Exercises: Choose a 1D, 2D, or 3D case; evaluate RMSE and discuss optimization trade-offs.
Format
- 5 live sessions (2 hrs each) + labs
- Office hours with instructors
- Capstone on your domain
- Certificate upon completion
Next Cohort
23-27th Feb 2026
Time
TBD
Mode
Virtual
Instructors
Prashant Srivastava - Co-Founder & CTO
PhD (Computational Mechanics). 10+ yrs in multiphysics modeling. Built production PINN pipelines for batteries, thermal systems, and structures.
Prajakta Sabnis - Co Founder & COO
M Tech, Mircoelectronics IIT Bombay, driving customer results and relationships.
Apply for the next cohort
Enterprise workshops
Custom programs for teams: domain‑specific curricula, lab data integration, co‑development of a pilot model, and deployment guidance.
Contact usTeam size
8–30
Duration
2–5 days