I am a Software Engineer and recent Master of Science in Computer Science graduate from UCM, bridging the gap between rigorous systems engineering and cutting-edge artificial intelligence. I specialize in building complex, high-performance architectures—from low-latency mobile/desktop applications to full-scale autonomous agentic backends.

Beyond applied engineering, I have a profound, relentless passion for the internal mechanics of deep learning. Rather than treating AI as a black-box API, I am committed to exploring neural network arithmetic, Transformer forward pass mechanics, and modern inference optimizations. I believe that true innovation stems from fundamentally understanding the mathematics and hardware execution (e.g., Tensor Cores, memory-bandwidth walls) that power intelligent systems.

My engineering philosophy focuses on:

• 🤖 Deep AI Architectures: Exploring Hybrid models (Mamba/SSMs), Sparse Mixture of Experts (MoE), and Multi-Token Prediction (MTP).
• 🧠 Agentic Orchestration: Designing both deterministic and non-deterministic agent meshes (Plan-and-Execute, Dynamic Supervisors) utilizing the Model Context Protocol (MCP).
• ⚡ Resilient Systems Design: Developing zero-latency, local-first applications with robust offline synchronization.
• 📐 Enterprise-Grade Engineering: Implementing Clean Architecture, MVVM, and secure cryptographic storage (PBKDF2-derived access).

My commitment to AI extends into continuous self-driven exploration of low-level architecture and mathematics. I am constantly studying how 2026-era optimizations alter the computational landscape:

• Mathematical Inference Modeling: Deconstructing the Transformer forward pass, attention scoring, and Softmax implementations.
• Hardware-Level Optimization: Studying KV Caching, FlashAttention-4 implementations on Blackwell architectures, and the shift from $O(N^2)$ to Linear-Scaling alternatives.
• Hybrid Architectures: Exploring the fusion of Self-Attention with State Space Models (Mamba/Jamba) for infinite context processing via Ring Attention.
• Agent Orchestration: Exploring scalable orchestration patterns from sequential pipelines to Magentic (Dynamic Supervisor) systems governed by strict JSON Schema validations.

Your personal, multi-platform AI assistant built with a modular, extensible architecture. Tech Stack: Python • FastAPI • TypeScript • Tauri • React • Qdrant (Vector DB) • Redis

• Autonomous Agent Core: Engineered background job processing with ARQ, real-time SSE streaming, and an event bus architecture.
• Advanced Integrations: Built 29+ REST endpoints, integrating Ollama, Mem0 memory layer, and DuckDuckGo web research.
• System Telemetry: Implemented low-level system monitoring tools tracking CPU, memory, and deep Windows Event Logs.

A futuristic desktop-first scheduling application featuring energy-aware intelligence. Tech Stack: Flutter 3.x • Dart • Drift ORM • SQLite • Clean Architecture

• Energy-Aware Intelligence: Designed algorithms mapping task scheduling to cognitive peak hours for productivity optimization.
• Architectural Rigor: Built on standard Top 1% codebase structures—employing Repository patterns, 7-day rolling windows, and extensive documentation frameworks.
• Polished UI/UX: Shipped a fully responsive, cross-platform glassmorphic UI with dynamic floating Focus Modes.

The ultimate offline-first Android media player leveraging the latest native technologies. Tech Stack: Kotlin (100%) • Jetpack Compose • Media3 (ExoPlayer) • Room DB • Dagger Hilt

• Advanced Playback Engine: Implemented seamless multi-track support, Picture-in-Picture (PiP), and sophisticated gesture controls.
• Reactive Architecture: Utilized rigorous MVVM with StateFlow, seamlessly integrating background Media3 playback services.
• Analytics Engine: Developed a robust local-first analytics dashboard capturing activity streaks and real-time listening habits.

A secure personal journaling application with cryptographic guarantees. Tech Stack: Flutter • Dart • ObjectBox • Cryptography (PBKDF2)

• Zero-Knowledge Security: Engineered XOR encryption for journal content paired with PBKDF2-derived key management (100,000 iterations).
• Hardened Authentication: Integrated hardware-backed keystore/keychain access, biometric fallbacks, and strict rate-limiting protocol against brute forcing.
• Resilient Data Integrity: Implemented 3-second delayed auto-saves, draft recovery buffers, and fully encrypted exports.

Actively contributed to the OpenClaw ecosystem, an advanced agentic framework bridging the gap between LLM inference and local execution environments.

• Core Architecture: Developed within the core plugin API boundaries, Gateway node wire protocols, and execution runner contexts.
• Enterprise Standards: Operated under strict CI/CD guidelines, high-coverage Vitest guardrails, and secure OSS supply chain rules (GHSA).

Progressive web application integrating AI intelligence into version control workflows.

• Intelligent Tracking: Vector-based context retrieval evaluating repository state and code dynamics to summarize development workflows.
• Tech Stack: TypeScript • React • Node.js • Prisma • SQLite (Vector)

Theory: Transformer Math, Mamba/SSM Hybrids, FlashAttention-4, MoE Routing, KV Caching
Agents: LangChain, MCP (Model Context Protocol), Magentic Supervisors
Infra: Qdrant (Vector DB), Redis, Ollama, FastAPI, Python

Android: Kotlin, Jetpack Compose, AndroidX Media3, Room, Dagger Hilt
Cross-Platform: Flutter, Dart 3.0+, Drift ORM, ObjectBox, Tauri
Focus: Offline-First Performance • Cryptographic Storage • Clean Architecture

Stack: TypeScript, React, Prisma, C#, ASP.NET, SQLite, Node.js
Focus: API Design • Real-Time Systems (SSE/WebSockets) • Progressive Web Apps

I am continuously exploring the intersection of Deep Learning mathematics, Autonomous Agent behaviors, and Elite Software Systems, bringing an unyielding commitment to contributing to AI.

_{Driven by mathematical curiosity and extreme engineering precision.}