// hello, world
I'm Aayush Randeep
Computational Cosmology · High Performance Computing · Scientific Software
About Me
Simulating the Universe,
One Grid Cell at a Time
I'm a final-year BS-MS student at the Indian Institute of Science Education and Research, Bhopal, majoring in Physics with a CPI of 9.11/10.
My research lives at the intersection of computational cosmology, high performance computing, and scientific software development for cosmological simulations. I build and optimize solvers for PDEs arising in physics simulations — targeting performance at the hardware level through techniques like compute culling, cache-aware scheduling, multigrid acceleration, and adaptive domain decomposition.
I'm equally drawn to theoretical cosmology — from modified inflationary models to lattice simulations — and to scientific computing as engineering: writing GPU-accelerated solvers, memory-efficient algorithms, and parallel code that scales on heterogeneous architectures.
Research Experience
Where I've Been Working
- Developed high-performance C++ solvers for inflationary perturbation equations, implementing the Mukhanov-Sasaki formalism with adaptive e-fold sampling and Bunch-Davies initial conditions to extract the primordial power spectrum Pζ(k)
- Built a NMDC Model of inflation with Ricci Tensor as the Curvature Tensor
- Formulated a consistent expansion scheme to incorporate higher-order curvature-induced terms in generalized NMDC models, preserving predictive stability of inflationary dynamics
- Constrained model parameters using ACT likelihood data and evaluated consistency with observational bounds on ns and r
- "Testing the NMDC models of inflation in the light of the latest ACT observations" — Draft completed
- Studying the resulting power spectrum in modified inflation models through HPC simulations in beyond slow roll setting
- 8-week onsite internship on the Param Shivay Supercomputer for simulating Solar Dynamo
- Rewrote STABLE code for Sun-like stars; identified stable radius regime for solar dynamo behavior while showing negligible dependence on time period
- Worked on Surya Code for simulating 2-D Solar Dynamo incorporating Lambda Effect
- Tools: FORTRAN, Linux Bash, HPC cluster
- Hybrid internship (4 weeks onsite + 8 weeks remote) in X-Ray Astronomy and astronomical datasets
- Performed AGN Data Analysis on eROSITA observations
- Validated LADE model predictions against eROSITA AGN observations, finding near-perfect agreement between theoretical distribution and observational data
- Tools: Python, Astropy, HEALPix, Colossus, SciPy, NumPy
- Studied fundamentals of Radio Astronomy
- Learned Astropy applications for analyzing SDSS Data
Projects
Things I've Built
Game-Engine-Inspired Navier-Stokes Solver
Solving PDEs on memory-constrained systems by emulating game engine resource management. Implemented 10 hardware-aware optimizations including multigrid V-cycles, adaptive compute culling (92.2% tiles skipped, 97% work saved), cache-friendly refetching, delta-encoded halo compression, and preallocated memory pools (0 runtime mallocs). GPU acceleration in active development targeting ROCm/HIP.
Neural-Numerical: Hybrid ML Root Finder
Hybrid solver using a neural network to provide warm-start initial guesses for Newton-Raphson, eliminating convergence failures. Working on making it suitable for stiff functions.
N-Dimensional Wave Equation Analysis
Studied PDEs, Real Analysis and Multivariable Calculus in N dimensions. Developed intuition for qualitative PDE analysis — extracting physical properties of wave systems without explicit solution.
Technical Skills
Tools of the Trade
Programming Languages
Scientific & HPC Tools
Systems & Miscellaneous
Education
Academic Background
Indian Institute of Science Education and Research, Bhopal
BS-MS in Physics — Final Year
December 2021 – Present
Cumulative Performance Index: 9.11 / 10
Thesis: Modified Inflationary Models & HPC Lattice Simulations
Relevant Coursework
Publications & Presentations
Work in Progress
"Testing the NMDC models of inflation in the light of the latest ACT observations"
Built a NMDC model of inflation with Ricci Tensor coupling and constrained model parameters using ACT likelihood data. Evaluated consistency with observational bounds on ns and r.
"Cosmic Inflation: The Story of the Beginning of Everything"
Presented at the In-House Physics Symposium 2026. Covered NMDC inflation with Ricci tensor coupling; results shown to be consistent with ACT DR6 observations.
Let's Connect
I'm actively seeking PhD positions in computational cosmology and roles in scientific computing & software development. Feel free to reach out — I'd love to hear about opportunities.