Profile & Vision
Attila Joseph Farkas
Lead R&D Engineer • Ph.D. Experimental Psychology
CNS • I/O • AI • APP
I build production-grade AI architectures at the intersection of human cognition and software engineering.
My background spans computer science, cognitive neuropsychology, and 3D motion graphics. I don't just research concepts; I build and commercialize them. A key example is my work engineering an internationally patented technology—System and Method for Diagnosing and Assessing Therapeutic Efficacy of Mental Disorders (PCT/US17/51345)—which I drove from foundational base research through experimental software design and the complex patenting procedure.
Today, my focus is bridging the gap between stochastic AI behavior and rigorous production standards. I build specialized cognitive runtimes that translate natural language-based rules into deterministic, Boolean-based smart functions. This approach offloads large language models from executing complex policy decisions and entirely removes them from the runtime critical path.
By designing custom Model Context Protocol (MCP) ecosystems and advanced AI Red Teaming (AIRTP+) frameworks, I engineer functional defensive guardrails directly into local workflows. The result is a formal neuro-symbolic framework that systematically constrains non-deterministic outputs, prevents AI slop, and ensures every cognitive step is mathematically verifiable and secure.
Selected Work & Background
AI Systems Architect
Rutgers University Office of Enterprise Application Services•Full-time•Hybrid
- Secure Local AI Infrastructure: Architected and deployed a custom VS Code extension utilizing a Model Context Protocol (MCP) server. This ecosystem enables large language models (like Gemini) to interact directly and securely with local filesystems and private APIs.
- Enterprise Cost Optimization: Engineered a local-first operational architecture that effectively replaces standard enterprise Copilot subscriptions, dramatically reducing cloud compute costs while maintaining high-fidelity algorithmic assistance.
- AIRTP+ Security Gating: Leveraged advanced AI Red Teaming (AIRTP+) methodologies to implement strict defensive guardrails directly within the MCP server. This deterministic gating systematically neutralizes prompt injection attacks and prevents unauthorized tool execution in production-grade AI agents.
ReactNext.jsTypeScriptAI Integration
(UI) Software Developer
TEKsystems, ETS•Contract
- Enterprise Frontend Architecture: Engineered type-safe, highly scalable user interfaces for large-scale assessment platforms, leveraging strict TypeScript and Redux.js for rigorous global state management.
- Custom Component Engineering: Architected and optimized advanced functional wrappers for complex third-party libraries (React-Uploady, Redactor) utilizing styled-components. Enforced strict UI consistency and automated testing across the organization using Storybook and Figma design systems.
- Secure Data & State Pipelines: Seamlessly bridged front-end architectures with robust backend services, building highly efficient data-fetching pipelines via RTK-Query and implementing secure OAuth2 authentication protocols within single-page applications.
TypeScriptRedux.jsUI/UX
Independent Full-Stack Developer & Digital Designer
Tau R&D•Self-employed•New Jersey
- Full-Stack Platform Engineering: Architected and engineered a comprehensive real estate platform prototype from the ground up, utilizing React and Redux for complex state management backed by a flexible MongoDB data architecture.
- Immersive 3D Integration: Bridged the gap between traditional UI and immersive environments by developing interactive 3D visualizations and custom motion graphics, elevating the user experience and securing stakeholder alignment.
- Rapid Front-End Architecture: Designed modular, highly responsive front-end interfaces using a hybrid Tailwind CSS and Bootstrap approach to ensure rapid iteration and pixel-perfect rendering across all devices.
ReactReduxMongoDB3D GraphicsMotion Graphics
Independent R&D Engineer (Contract)
AUFHAVEN PTE. LTD•Freelance•Singapore
- Brain-Computer Interface (BCI) Architecture: Directed the software engineering for mobile BCI prototypes, integrating React frontends with Unity 3D engines to capture and process live data from wearable EEG headsets.
- Low-Latency Neural Telemetry: Engineered high-speed, real-time data pipelines utilizing the Open Sound Control (OSC) protocol, ensuring the low-latency transmission of complex neural signals.
- Real-Time Data Rendering: Translated raw, high-frequency brainwave telemetry into dynamic visual feedback loops, seamlessly bridging raw hardware sensor data with responsive, user-facing digital environments.
ReactUnityOSCEEG Visualization3D UI
Contract R&D Scientist (Digital Health)
Janssen Research & Development, LLC•Contract
- Gamified Diagnostic Architecture: Directed the experimental design and software architecture for a novel diagnostic prototype, engineering an interactive digital system specifically designed to detect elevated signs of suicidal ideation.
- Clinical-to-Code Translation: Bridged the gap between clinical psychology and mobile engineering by successfully translating analog, evidence-based psychological assessments into dynamic, highly functional React and Android applications.
- Secure Data Integration: Managed the underlying data architecture for the research trials, structuring secure pipelines to capture and quantify high-stakes mental health metrics for enterprise-level R&D analysis.
ReactAndroid StudioClinical R&DPrototyping
Product Architect & R&D Engineer
Multichannel Haptic Wearable (Startup)•Fellowship
- End-to-End MVP Engineering: Directed the complete product lifecycle for a novel neck-wearable navigational aid, designing both the 3D-printed physical hardware enclosure and the underlying software control architecture from scratch.
- Sensory Bottleneck Mapping: Leveraged my background in cognitive neuropsychology to conduct foundational research into human haptic systems, systematically identifying perceptual bottlenecks to ensure the device matched human processing limits.
- Signal Optimization & User Testing: Developed and rigorously user-tested the tactical feedback algorithms, optimizing the multichannel array to achieve maximum signal distinction and creating a highly intuitive, zero-screen navigational interface.
Lean Startup3D PrintingHaptic FeedbackUser Testing
Inventor & Lead R&D Engineer
Rutgers University, Laboratory of Vision Research•Contract
- End-to-End Product Architecture: Architected and developed a complete diagnostic software system designed to assess the therapeutic efficacy of mental disorders, transitioning the project from foundational theoretical research into a functional, experimental software prototype.
- Commercialization & IP Strategy: Navigated the complex technology transfer and research commercialization pipeline, successfully securing an international patent-pending status for the system and underlying methodology.
- Clinical Data Integration: Engineered the system to process and quantify complex psychological and biometric data, establishing the deterministic baselines required for clinical-grade technological tools.
AndroidMATLABPython3D CGIPatents
Core Tech Capabilities
Curated skills that enable real-world AI integration
Agent Building
LangGraph / AutoGenCognitive ArchitecturesLLM Fine-tuningVector Databases (Pinecone, Chroma)RAG Pipelines
Coding & Systems
Flutter (Gemma integration)TypeScript / Next.jsPython (PyTorch, Pandas)React (rtk query)Docker / CI/CD Pipelines
Scientific Integration
Psychometric data & 3D CGIPsychoacoustic data & Headphone designAI integration into cloud serviceLocal AI integration into Flutter App