Structural Maturation of Neural Networks & Subjective Time
Investigates the relationship between the structural maturation of complex neural networks — modeled with Liquid State Machines — and temporal perception.
Applied R&D for complex real-world problems.
We help organizations turn difficult scientific and technological questions into models, methods, experiments and validated prototypes.
From complex science to real-world solutions.
Applied R&D
When the solution is not yet obvious
Some problems cannot be solved by selecting a tool or implementing a known recipe. They need to be investigated.
ComplexitySoft acts as an external R&D capability for organizations that need to understand, model, test and validate difficult scientific or technological problems before committing to a larger implementation.
An external R&D capability when building one internally is impractical, premature or insufficient.
-
No internal R&D?
We can structure the investigation from the beginning.
-
Limited R&D capacity?
We can extend your team with specialized methods and research capability.
-
Established R&D?
We can collaborate on focused problems that require external expertise or independent validation.
Typical outputs
How we work
From problem formulation to evidence-backed prototypes.
- 01
Frame
Clarify the problem, constraints, hypotheses and success criteria.
- 02
Investigate
Review evidence, data, methods and technical feasibility.
- 03
Model
Formalize the problem through computational, statistical, experimental or mathematical models.
- 04
Build
Develop algorithms, analyses, simulations, PoCs or functional prototypes.
- 05
Validate
Test assumptions, quantify results, assess robustness and document limitations.
- 06
Transfer
Deliver reproducible methods, code, documentation, findings and a roadmap for the next stage.
Not every project needs all six stages. We can join at any point — and the difference is greatest when we are involved from the start.
Ways to engage
-
Full-cycle R&D
From problem formulation through investigation, modeling, prototyping, validation and transfer.
-
Focused engagements
A single stage when that is what the problem needs — such as problem framing, technical feasibility, modeling and simulation, experimental design, proof of concept or validation.
-
Collaborative R&D
Specialized expertise, independent investigation, methodological support or additional research capacity alongside your researchers and engineers.
Expertise
Problem-led and method-driven
We work across domains where scientific reasoning, computational modeling, data and intelligent systems can unlock a solution.
- 01
Computational Modeling & Scientific Analysis
Models, simulations and quantitative methods for turning poorly structured questions into testable representations.
- 02
AI & Intelligent Systems
Applied machine learning, model development and evaluation, and intelligent-system prototypes for research-oriented technical problems.
- 03
Complex Systems & Network Science
Complex networks and graph-based methods for understanding interacting systems, emergent dynamics, structural patterns and outliers.
- 04
Human Perception & Behavioral Science
Psychophysics, eye tracking, experimental design and behavioral-data analysis for problems involving human perception and behavior.
- 05
Agentic Architectures
Agent workflow design, agentic system architecture and orchestration concepts, explored through prototypes and proofs of concept.
Enabling capabilities
Scientific software Python Statistical computing Data processing Visualization Computer vision Cloud & databases when required
Selected research
Research that expands what we can build
Our research activity is not separate from applied R&D. It is how we deepen methods, test ideas and develop capabilities before they become obvious solutions.
Visual Information Availability & Facial Exploration
Examines how the availability of visual information relates to gaze behavior and the hierarchical organization of facial exploration in emotion recognition, combining eye tracking and psychophysics.
Complex-Network Methods for Classification & Outlier Detection
A line of research using complex-network representations and graph-based analysis for classification and for detecting structural outliers, including community detection in multilayer networks.
Team
People
Complex problems often live between disciplines. Our work combines computational and human-science perspectives.
Founder | AI Research & Strategy
Rafael Delalibera Rodrigues, PhD
PhD & MSc · University of São Paulo (USP)
Computer scientist and data scientist with a PhD in Computer Science and Computational Mathematics from the University of São Paulo. His work spans artificial intelligence, complex systems, network science, computational neuroscience, modeling and scientific computing, connecting research methods with applied technological solutions.
Scientific & Data Science Collaborator
Melina Boratto Urtado, PhD
PhD & MSc in Psychobiology · University of São Paulo (USP)
Psychologist, psychobiologist and data scientist with a PhD in Psychobiology from the University of São Paulo. Her work integrates psychophysics, visual perception, eye tracking, facial processing, experimental design and quantitative behavioral-data analysis, combining human-science methods with computational and statistical approaches.
About
Research-driven by design
ComplexitySoft is built around a simple idea: difficult problems deserve more than off-the-shelf answers.
We combine scientific reasoning, computational methods and rapid prototyping to investigate questions before scaling solutions. Alongside client R&D, ComplexitySoft Research maintains an active internal research track that develops methods and scientific capabilities for future applied work.
Research expands capability; Applied R&D converts capability into real-world value.
Contact
Have a problem that needs investigation before implementation?
Tell us what your organization is trying to understand, model, test or build.
Research & scientific collaboration: research@complexitysoft.com