Description

Software Engineer & ML Developer specializing in Computer Vision and NLP applications. I build production-ready AI systems that solve real-world problems.

🔹 Expertise:

- Computer Vision: YOLOv11, object detection, image segmentation (93.2% mAP achieved)

- NLP & Translation: Transformer models, context-aware translation systems

- Full-Stack ML: Python, PyTorch, TensorFlow, FastAPI, Flutter

- MLOps: Docker, CI/CD pipelines, model deployment, Azure

🔹 Recent Projects:

- WasteCareAI: AI waste classification system with 5,300+ annotated images, real-time detection (>15 FPS)

- SpeakAndTranslate: Cross-platform language learning app with Gemini 2.0 integration

🔹 Technical Approach:

I follow clean architecture principles, implement comprehensive testing (pytest, JUnit), and maintain detailed technical documentation. My solutions are scalable, well-documented, and production-ready.

Languages

Spanish
Native or bilingual
English
Fluent
German
Conversational

Workplace preferences

Remote only

Primarily works remotely

SpeakAndTranslate
Software Engineer & ML Developer
EDUCATION AND E-LEARNING
January 2025 - Today (1 year and 6 months)
Bogotá, Colombia
Development of a cross-platform translation and language learning app with Flutter and a Python Backend (FastAPI). The application supports real-time translation, voice commands, natural speech synthesis, and word-for-word learning features.

- Development of mobile frontend with Flutter (Android, iOS, Web) including chat interface, audio recording, audio playback, and text-to-speech mode
- Implementation of Python backend with FastAPI for translation, text-to-speech, speech-to-text, and audio processing
- Integration of Gemini 2.0 for context-based translations and four different translation styles (native, colloquial, informal, formal)
- Use of Azure Cognitive Services Speech for high-precision language recognition and natural speech synthesis
- Development of a word-for-word translation system with authentic pronunciation in Spanish, German, and English for learning purposes
- Architecture following Clean-Architecture-Principles (data, domain, and presentation layers)
- Backend containerization with Docker and deployment via Azure Container Apps
- Automated CI/CD pipelines with GitHub Actions (build, Docker push, deployment)
- Creation of technical documentation, API specifications, and architecture diagrams

GitHub: https://github.com/codemotozu/Documented_Speak_and_Translate_App
Demo: https://codemotozu.github.io/speak_and_translate_app/
Natural Language Processing Flutter FastAPI Microsoft Azure Python
WasteCare.AI
Data Scientist & ML Engineer
ENVIRONMENTAL
January 2024 - August 2025 (1 year and 7 months)
Bogotá, Colombia
Development of an AI-powered Computer Vision system for automated classification of household waste using Ultralytics YOLOv11 (PyTorch) and instance segmentation. Creation and processing of a specialized dataset with 5,300 annotated images and 10 waste categories, including comprehensive data augmentation.

- Construction and management of the dataset in Roboflow: Annotation, quality control and data processing
- Training and evaluation of the YOLOv11 model with PyTorch, validation strategies and hyperparameter optimization
- Achieved mAP@50 of 93.2%
- Implementation of a real-time video processor (>15 FPS) for object detection, segmentation and classification in live streams
- Application of analytical methods to investigate data distribution and quality (Pandas, NumPy, Matplotlib)
- Integration and testing of the Segment Anything Model (SAM, PyTorch) for additional segmentation tasks
- Creation of a 60-page scientific documentation with data analysis, architecture diagrams and evaluation metrics

Documentation (GitHub): https://codemotozu.github.io/WasteCareAI_PDF
Live-Demo: https://codemotozu.github.io/waste_care.ai/
Computer Vision Pytorch Python Machine learning Deep Learning