the work · the engineer
I find the order beneath the noise.
ML engineer & CAE-AI lead in Darmstadt. I build models and agents that find the hidden structure inside three-dimensional crash simulations — anomaly detection, generative geometry, root-cause analysis. I came up through mechanical engineering and a love of leading teams; still mid-service, building toward founding.
- CAE-AI
- anomaly detection
- generative geometry
- agentic AI
- leadership
the timeline · the record, in order
One machine, a life’s work.
The same car — assembled, meshed, analysed, broken, and read for the one fault that matters. Education and career, in the order they happened. Scroll.
I built things before I modelled them.
B.E. Mechanical Engineering · RMK Engineering College (Anna University)
GPA 8.9/10 · two-time best of class. Captain of a 14-engineer EcoKart team — All-India Rank 13 and the Best Innovation Award, 2015.
Then I learned to interrogate them.
Project Engineer · CAE — Wipro Technologies
Automated the sub-assembly crash-simulation setup & review in ANSA (Python) for a Japanese OEM. Best Newcomer Award, 2017.
I traded the wrench for the model.
M.Sc. Computer-Aided Mechanical Engineering (CAME) · RWTH Aachen
Two-time best of class · Dean’s List (top 5%). Carved a path from mechanical engineering into AI for computational science.
I taught a process to optimise itself.
Student Research Assistant · Cybernetics Lab (IMA & IfU), RWTH
Reinforcement learning to optimise the hot-rolling of steel; a FastAPI service logging every run into an ArangoDB graph database.
I taught a model to read the crash.
Master’s Thesis · Research — Porsche AG
Few-shot, self-supervised learning on crash meshes — geometric deep learning with PointNet++ autoencoders to flag critical deformation.
Now I find the one that doesn’t belong.
ML Engineer · CAE-AI Lead — Kube GmbH
Anomaly detection on 3D crash simulations in production for Porsche; a GPT that generates mesh from CAD; published in the Int. Journal of Crashworthiness.
03 · the ledger · selected work
The work, as cinema.
Production CAE-AI first, then the research it grew from, then where it began. Scroll — the reel runs sideways. Each frame opens its file.
Outliers in the Crash
Teaching a model to flag the simulation that does not belong — the outlier crash among thousands.
A Grammar for Geometry
A GPT that generates mesh from CAD — a language model learning the grammar of three-dimensional form.
Learning from Few Crashes
Detecting critical deformation when labelled examples are scarce — data-efficient learning on crash studies.
The Intelligent Finite Element
Predicting stress fields with a neural network — finite element analysis at a fraction of the compute.
Rolling Steel by Reinforcement
Optimizing the hot-rolling of steel with RL on Gym-compatible simulations — and the API to log it all.
The Incremental Tour Planner
A sequence-to-one policy that builds an optimized route in real time, adapting as the problem changes.
Denoising the Signal
A fully convolutional autoencoder that cleans accelerometer signals better than a classical filter.
Reading a Walk
Distinguishing normal from impaired gait from phone sensors — a small tool with a human purpose.
Where the Robot Is
A compact SLAM implementation — localising a robot in a 2D world from its motions and senses.
An Electric Kart, From Scratch
Captain of fourteen engineers. We built an electric kart from nothing and placed AIR 13.
swipe sideways · each frame opens its file
04 · the instrument · capabilities
The shape of what I know.
A live network — areas and the tools inside them, each sized and scored by how strong I run there. Hover any node for its card.
drag the nodes · hover any node for its card
A capability map.
Each node is a skill — bigger and brighter where I run stronger, the number is the score. Tools are sub-nodes with their own scores. Hover or drag any node for its card.
- CAE Analysis95
- Machine & Deep Learning92
- Programming90
- Agentic & Generative AI88
- CAD / CAE Automation86
- MLOps / Cloud74
- Data / Graph76
05 · the marks · recognition
Marks left along the way.
recognition
Dean's List
Top-5% of the degree program.
Best of Class Awardee
RWTH Intac scholarship — academics + extracurriculars.
Best of Class Awardee
RWTH Intac scholarship — first semester.
Best Newcomer Award
Wipro — automating crash simulation setup in ANSA.
Best Innovation Award
Eco-Kart — fire extinguisher, IR detection, smart belt.
published
International Journal of Crashworthiness
2022
Method for Automated Detection of Outliers in Crash Simulations
Co-authored with Dr. David Kracker.
DOI 10.1080/13588265.2022.2074634
Int. Journal for Research in Applied Science & Engineering Technology
—
Automatic Fire Extinguisher
Best Paper Award · with Mr. Karthick Selva.
Still mid-service — and the next posting hasn’t converged.
If you’re building something that needs an engineer who reads the physical world through models, let’s talk.