• 📐 Mathematics senior student at Universidad Yachay Tech (Ecuador).
• 🤖 Machine Learning Engineer trained via Anyone AI & Udacity.
• ☁️ Cloud Application Developer candidate at AWS Cloud Institute.
• 🎯 Research & Engineering Focus: Scientific Machine Learning (PINNs), Computer Vision for Bio-Imaging, Transformers & NLP, and Applied Statistics.

• SentimentScope: Transformer NLP Architecture
  Modular 4-layer Transformer encoder built from scratch in PyTorch (multi-head self-attention, positional encoding, custom embeddings). Trained with AdamW on GPU with 100% test pass rate CI/CD.
  PyTorch Hugging Face Pytest CI/CD • Project Dossier

• CheX-DS: Pneumonia Detection with Transformers
  Hybrid architecture (DenseNet-121 + Swin Transformer) for pediatric chest X-ray screening. Achieved 99.0% recall with Weighted Asymmetric Loss and Grad-CAM interpretability.
  PyTorch Swin Transformer Streamlit Docker • Project Dossier

• PINNs & SciML Laboratory
  Continuous operator learning framework solving 1D Burgers, Heat Diffusion, and Navier-Stokes equations with physics conservation loss, automatic differentiation, and AWS S3/EC2 deployment.
  PyTorch DeepXDE FastAPI Docker AWS • Project Dossier

• 3D Isotropic Bio-Image Reconstruction
  Unpaired image-to-image translation with 3D CycleGAN in PyTorch to resolve axial point-spread-function (PSF) distortion in confocal fluorescence microscopy.
  PyTorch CycleGAN Bio-Imaging LaTeX • Project Dossier

• Energy Forecast Hub
  Production-grade time-series forecasting microservice with XGBoost and Random Forest, orchestrated via FastAPI and interactive Streamlit UI.
  Python FastAPI Streamlit Docker XGBoost • Project Dossier

A computational laboratory dedicated to interactive numerical simulations and theoretical foundations on jeffersonconza.github.io/matematica-ia:

• 🎯 Continuous Dynamic Programming: Value function derivation and policy convergence in Bellman optimality.
• 🔠 QKV Multi-Head Attention: Vectorized scaled dot-product attention simulator in latent projection spaces.
• 📐 17 Stewart Equations & Millennium Problems: Rigorous mathematical breakdowns of fundamental laws and open conjectures.

import torch

def physics_loss(u_net, x, t, nu=0.01 / torch.pi):
    """
    1D Viscous Burgers' equation residual:
    f := ∂u/∂t + u (∂u/∂x) - ν (∂²u/∂x²) = 0
    """
    x.requires_grad_(True)
    t.requires_grad_(True)
    
    u = u_net(torch.cat([x, t], dim=1))
    
    # Exact differential operators computed via PyTorch automatic differentiation
    u_t = torch.autograd.grad(u, t, grad_outputs=torch.ones_like(u), create_graph=True)[0]
    u_x = torch.autograd.grad(u, x, grad_outputs=torch.ones_like(u), create_graph=True)[0]
    u_xx = torch.autograd.grad(u_x, x, grad_outputs=torch.ones_like(u_x), create_graph=True)[0]
    
    residual = u_t + u * u_x - nu * u_xx
    return torch.mean(residual ** 2)

• Languages: Python, R, Julia, SQL, C++, Bash, LaTeX
• Deep Learning & SciML: PyTorch, DeepXDE, Hugging Face, Torchvision, TensorFlow, scikit-learn
• Scientific Computing: NumPy, SciPy, pandas, SymPy, Matplotlib, Manim
• Cloud & Deployment: Docker, AWS (EC2, S3, SageMaker), FastAPI, Streamlit, Git, Linux

• Cosmos: Alexander von Humboldt y la Red de la Vida
• Telescopio Espacial Nancy Grace Roman & Energía Oscura
• PINNs: Redes Neuronales Informadas por la Física
• Puntos de Lagrange: Mecánica Celeste y Equilibrio Orbital

Explore all bilingual essays and notes at jeffersonconza.github.io/blog.