[ murat.khidoyatov ]

$ cat identity.txt
→ Mechanical Engineer × Computational Researcher
$ grep --institution
→ Yale University · B.S. Mechanical Engineering '27
$ ls ./focus/
CFD/  PINNs/  GPU-acceleration/  aerospace/  mechanism-design/
$ _

Engineering the things that don't exist yet.

// location New Haven, CT

// status Open to new projects and opportunities

// contact murat.khidoyatov@yale.edu

About

I'm an undergraduate at Yale University, studying mechanical engineering with a strong pull toward computation. I work at the intersection of physics-based modeling, machine learning, and aerospace systems — building things that need to move, deform, and survive in the real world.

My research lives in the Humphrey MechanoBiology Lab, where I develop physics-informed neural networks for hyperelastic materials and accelerate their training on NVIDIA L4 / A100 GPUs. Last summer I built ML-driven alloy optimization pipelines at Oak Ridge National Lab.

Outside the lab, I lead the Yale Undergraduate Robotics team, where we're designing a Mars rover life-detection system from scratch. I also lead a Yale - Assistive Independent Design (AID) team building custom pediatric prosthetic limbs for local patients. I love creating tools and systems that address real and novel problems, and I'm always looking for new challenges to tackle.

Research & Publications

ongoing 2023 — Present

Physics-Informed Neural Networks for Hyperelastic Mechanics

Yale Engineering MechanoBiology Lab · Undergraduate Researcher

Developing a JAX-based dual-network architecture that couples displacement prediction with a learned energy function. Uses automatic differentiation to enforce governing PDEs as soft constraints, achieving 95% accuracy relative to finite element solutions on benchmark deformation problems. GPU-accelerated training reduced iteration time on L4 / A100 hardware.

JAXPyTorchCUDACFDPINNsNeo-Hookean

internship Summer 2025

ML-Driven Alloy Design under Sparse Experimental Data

Oak Ridge National Laboratory · Research Intern

Built ML surrogate models (96% predictive accuracy) integrated with Bayesian Optimization to navigate high-dimensional alloy design spaces. Combined surrogate inference with acquisition functions to efficiently explore noisy datasets near global optima. Validated physical interpretability through feature-response correlation analysis.

PythonBayesian OptimizationRandom ForestMaterials

// publication · pending submission

Shirani, M., Gueldner, P. H., Khidoyatov, M., Warren, J. L., Ninno, F., & Humphrey, J. D. (2026). Physics-Consistent Neural Networks for Learning Deformation and Director Fields in Microstructured Media with Loss-Based Validation Criteria.

Projects

P-01

Mars Rover Life-Detection System

Lead Engineer · Yale Undergraduate Robotics

Leading mechanism design and structural integration of a deployable life-detection rover. Designed a 4-DoF robotic arm with a custom forward/inverse kinematic simulator for path planning, and validated a 50N linear actuator deployment mechanism to a 2.0× safety factor via FEA.

SolidWorksCOMSOLFEAPython

View project →

P-02

Pediatric Prosthetic Limb

Team Lead · Yale AID

Led a 7-member multidisciplinary team to deliver a custom passive prosthetic limb to a pediatric patient. Secured $5,000+ in project funding, performed root-cause analysis on pneumatic actuator failures, and iterated 3D-printed silicone mold designs for reliability.

SolidWorks FEADFA3D Printing

View project →

P-03

CFD-Trained PINN Pipeline

Research · Humphrey Lab

Developed Python CFD simulations of flow around a cylinder, generating ground-truth datasets used to train physics-informed neural networks. Accelerated training via CUDA-based GPU kernels on NVIDIA L4 / A100, enabling rapid iteration over large simulation datasets.

CFDJAXCUDANumPy

View project →

P-04

Custom Spectrometer

Yale Robotics · Science Sub-team

Designed and built a custom spectrometer from scratch using Fourier decomposition of light wave data and image analysis. Integrated with embedded sensor systems and custom sheet-metal extrusions.

Image AnalysisFourier MethodsSheet Metal

View project →

P-05

Nuclear Material Handling Tooling

Oak Ridge National Lab · R&D Intern

Designed and fabricated 3D-printed tooling for graphite handling using Design for Manufacturing principles, reducing process cycle time by 66% (12+ hr → 4 hr). Performed structural characterization on 100+ aluminum alloy samples.

DFM3D PrintingMaterials Testing

View project →

P-06

Constitutive Equation Discovery

Research · Humphrey Lab

Architected a deep learning model in JAX/PyTorch to automate constitutive equation discovery for elastomers. Achieved ±90% prediction accuracy on Neo-Hookean non-linear elasticity vs. COMSOL finite element solutions.

PyTorchJAXAuto-diff

View project →

Gallery

// physical builds, prototypes, lab work — keep scrolling

4-DOF rover arm test — 024-DOF rover arm validation

Oak Ridge research poster — 04ORNL — ML alloy optimization

Résumé

Two versions, tailored to different roles. Both PDFs.

→

RESUME · 01

Mechanical Engineering

Hardware-focused: CAD, FEA, mechanism design, fabrication, aerospace structures.

Download PDF

→

RESUME · 02

Machine Learning and Computational Simulations

Focus: ML, Scientific Computing, CFD, PINNs, GPU acceleration.

Download PDF