Scientific Computing & HPC Consulting

Accelerate your research & development with custom software solutions

Training Scientists provides specialized consulting services for scientific data analysis, interpretation, and visualization. We have a special focus in High Performance Computing (HPC) for the development and optimization of scientific simulations.

Schedule Free Consultation

Scientific Software Development

Algorithm Implementation

Transform research concepts into efficient, production-ready code

Data Analysis Pipelines

End-to-end solutions for processing large scientific datasets

Legacy Code Modernization

Update and optimize existing codebases for better performance and maintainability

Research Tool Development

Build specialized software tools for your specific scientific domain

High Performance Computing (HPC)

Performance Analysis

Profiling to identify bottlenecks and optimization opportunities

Parallelization

Implementation of MPI, OpenMP, and CUDA for CPU/GPU and hybrid computing

Scalability Solutions

Design systems that grow with your computational demands

Cloud HPC Integration

Integrate cloud computing resources with on-premise systems

Memory Optimization

Reduce memory usage and improve data access patterns

How We Work - Process Flow

How We Work

Initial Consultation

We start with a comprehensive discussion of your research goals, current challenges, and computational requirements.

Technical Assessment

We analyze your existing code, infrastructure, and workflows to identify optimization opportunities.

Solution Design

We develop a detailed proposal outlining our recommended approach, timeline, and expected performance improvements.

Implementation

We implement the solution using best practices, with regular progress updates and milestone reviews.

Knowledge Transfer

We provide comprehensive documentation, training sessions, and ongoing support to ensure your team can maintain the solution.

Success Stories

Biophysics Simulation Optimization

2,700x Speedup Achieved

Challenge

Biophysics simulation with 208 colloids taking 6 weeks to complete on a single core, severely limiting research capabilities and preventing larger-scale studies.

Solution

Implemented MPI parallelization and advanced algorithmic optimizations in C, designed for high-performance computing environments with thousands of cores.

Results

2,700x speedup - simulation with 64 times more colloids (13,312 total) now complete in just 24 hours using 4,096 cores.

208→13,312

Colloids Simulated

6 weeks→24h

Computation Time

1→4,096

CPU Cores

2,700x

Performance Gain

Visualization

In the image below, the evolution from the single core 208 squirmer to the parallelized 832 squirmer and the 3,328 squirmer simulations are shown, demonstrating the dramatic improvement in simulation capability.

Plasma Physics Code Development & Optimization

50x Performance Improvement

Challenge

Legacy Fortran code not adapted to 3D geometries, limiting simulation capabilities for complex stellarator configurations and preventing accurate modeling of modern fusion devices.

Solution

Refactored the entire codebase from 140,000 to 30,000 lines of code. Implemented 3D geometry handling and iterative linear equation solver for enhanced computational efficiency.

Results

Up to 50x performance improvement of the field solver, enhanced parallel scaling performance, enhanced maintainability, and added 3D visualizations.

140K→30K

Lines of Code

50x

Field Solver Speed

Geometry Support

Enhanced

Maintainability

Visualization

3D visualization of Wendelstein 7-X stellarator geometry shown below.

Additional Success Stories

Energy Trading Optimization Platform

Production-Ready Solution

Challenge

Complex energy storage simulation based on academic research equations was plagued with bugs and poor performance, hindering critical cost-benefit analysis for battery storage investments.

Solution

Developed Python simulation and interactive dashboard for real-time visualization of trading scenarios and profitability analysis with robust error handling and optimized algorithms.

Results

Delivered bug-free, production-ready code enabling rapid evaluation of energy storage investment strategies with real-time interactive dashboards.

100%

Bug-Free Code

Real-Time

Visualization

Interactive

Dashboard

Production

Ready

Visualization

Interactive dashboard displaying energy trading scenarios and battery storage profitability analysis, enabling stakeholders to make informed investment decisions through clear, real-time data visualization.

Postprocessing of Terabytes of Simulation Data

15x Performance Improvement

Challenge

Processing 10TB+ of simulation data with existing tools was taking weeks to complete, creating significant bottlenecks in research workflows and delaying critical analysis results.

Solution

Developed custom Python pipeline with parallel processing capabilities and efficient data structures, optimized for large-scale scientific data analysis and memory management.

Results

15x faster processing of massive datasets, transforming weeks-long computations into manageable timeframes and accelerating research discovery.

10TB+

Data Processed

Weeks→Days

Processing Time

15x

Speedup

Parallel

Processing

Visualization

Performance comparison showing the dramatic reduction in processing time for large-scale simulation data, demonstrating the efficiency gains achieved through optimized parallel processing algorithms.

Why Choose Training Scientists

Research-Focused Expertise

Deep understanding of scientific computing across multiple disciplines including physics, biology, and engineering.

Proven Track Record

Years of experience optimizing research codes and developing high-performance scientific applications.

Education + Implementation

Unique combination of teaching expertise and hands-on development skills ensures solutions that your team can understand and maintain.

Technology Agnostic

Expertise across multiple programming languages (Python, C, Fortran) and computing platforms (CPU, GPU, cloud, HPC clusters).

End-to-End Solutions

From initial concept to production deployment, we handle every aspect of your computational challenges.

Getting Started

Free Consultation

Contact us to discuss your computational challenges with our experts at no cost. We'll provide initial recommendations and help you understand how our services can benefit your research.

Custom Proposals

Every project is unique. We provide detailed, tailored proposals that outline scope, timeline, and expected outcomes specific to your needs.

Flexible Engagement Models

Project-based: Fixed scope and timeline for specific objectives

Ongoing support: Regular optimization and maintenance services

Emergency consulting: Rapid response for critical performance issues

FAQ (Frequently Asked Questions)

Do you work with existing codebases or only develop new solutions?

Both! We frequently optimize and modernize existing code. However, we also develop entirely new solutions from scratch.

What programming languages and platforms do you support?

Our primary expertise is in Python, C, and Fortran, with experience across HPC clusters, cloud platforms (AWS, Google Cloud), and GPU computing (CUDA). However, we can work with whatever technology stack best fits your existing infrastructure and requirements.

How do you ensure code quality and maintainability?

We follow industry best practices including documentation, version control, automated testing, and code review processes. All deliverables include detailed documentation and, when requested, training sessions to ensure your team can maintain and extend the solution.

Can you help with both short-term fixes and long-term projects?

Absolutely. We offer flexible engagement models from emergency performance troubleshooting (rapid response within 24-48 hours) to multi-month development projects. We can also provide ongoing optimization and maintenance support.

Do you provide training along with implementation?

Yes, knowledge transfer is a key part of our service. We provide documentation, code comments, and can include training sessions to ensure your team understands the implementation and can make future modifications independently.

How do you determine project costs and timelines?

Every project is unique, so we provide custom quotes based on scope, complexity, and timeline requirements. After our free initial consultation and technical assessment, we’ll provide a detailed proposal with fixed costs for defined deliverables or hourly rates for ongoing support.

About

Training Scientists is run by Dr. Maurice Maurer who graduated in 2020 from the Technical University of Munich with a Ph.D. in computational physics. In his scientific work – both carried out at the University of California, Los Angeles (UCLA) and the Max Planck Institute for Plasma Physics – he developed a high performance computing (HPC) code to simulate plasma microturbulence in complex geometries. The dissertation is called GENE-3D – a global gyrokinetic turbulence code for stellarators and perturbed tokamaks.

During his masters project he already developed a parallelized code running on some of the fastest supercomputers in germany. For both the masters project and the Ph.D. project he employed Python for the postprocessing of the large data sets generated by the simulations.

Also, he is doing Python development consulting work for for various companies in regards to code review and performance optimization.

The knowledge and experience from these projects in addition to more than 20 years of programming experience go into the Python programming courses.