About Me

Senior AI Software Developer

OpenSport / Playwize — Canada

• Built production-oriented AI agents and reasoning workflows for sports intelligence using LLMs and LangGraph.
• Designed multi-step agent workflows involving state management, tool selection, reflection, fallback handling, and error recovery.
• Developed RAG workflows using embeddings and vector search to ground LLM responses in domain-specific information.
• Integrated multiple LLM providers, including Anthropic, Google Gemini, and OpenAI, within provider-agnostic AI workflows.
• Implemented streaming AI responses through a Node.js gateway using Server-Sent Events (SSE).
• Integrated Langfuse for LLM observability, including tracing, prompt analysis, latency, and cost monitoring.
• Worked with BigQuery Vector Search and cloud-based AI infrastructure.

🔹 Focus: Generative AI · Agentic AI · LLMs · RAG · LangGraph · Vector Search · AI Observability · GCP

AI Engineer — Independent Contractor

Lisnen — Toronto, Canada

• Developed audio classification models using CNN architectures and pretrained audio representations, including YAMNet and PANNs.
• Built audio preprocessing and augmentation pipelines using Librosa and Audiomentations.
• Compared multiple model architectures and evaluated accuracy, generalization, and deployment constraints.
• Investigated TensorFlow Lite compatibility and optimized the inference approach for resource-constrained environments.

🔹 Focus: Audio AI · Deep Learning · TensorFlow · Keras · Model Evaluation · Edge Inference

AI Engineer — NLP Project

Angelina — Contract Project

• Developed a multi-class text classification pipeline using BERT to analyze large volumes of noisy and unstructured user feedback
• Experimented with both BERT and RoBERTa during model evaluation, with BERT selected for the final solution.
• Fine-tuned the Transformer model to capture semantic information from user feedback, addressing slang, spelling errors, noisy text, and significant class imbalance.
• Addressed common NLP challenges including slang, spelling errors, noisy text, and significant class imbalance in user feedback.
• Applied NLP techniques including Named Entity Recognition (NER) to identify products and services associated with negative feedback.
• Built data preparation and model training workflows using Python, Pandas, NumPy, PyTorch, and Hugging Face Transformers.
• Developed a production-ready inference pipeline with FastAPI, exposing REST endpoints for text classification and confidence scores.
• Containerized the inference service with Docker for deployment in the client's environment.

🔹 Focus: NLP · Transformers · Text Classification · Sentiment Analysis · Embeddings · NER · PyTorch · Hugging Face · FastAPI · Docker

Machine Learning Engineer Intern

BrainInnov — Montréal, Canada

• Developed 3D deep learning models for lung nodule classification using PyTorch and MONAI.
• Built preprocessing pipelines for 3D CT volumes using SimpleITK and MONAI.
• Experimented with ResNet, DenseNet, and attention-based architectures.
• Applied Grad-CAM to improve model interpretability and visual analysis of predictions.
• Evaluated models using accuracy, AUC, F1, precision, and recall.

🔹 Result: Achieved 87.5% accuracy and 0.91 AUC in the evaluated lung nodule classification model.

Python Developer / Data Engineer

Globant

• Developed Python-based data processing and ETL workflows for enterprise projects.
• Processed large Excel datasets using Pandas and built pipelines for data transformation and validation.
• Worked with AWS Glue, Athena, and S3 for cloud-based data processing and analytics.
• Contributed to data workflows supporting analytics and machine learning use cases.

🔹 Focus: Python · Pandas · AWS · Glue · Athena · S3 · Data Engineering

Full Stack Developer

Radix

• Developed backend and web applications using Python, Flask, JavaScript, and jQuery.
• Built data transformation workflows using PySpark and Databricks.

🔹 Focus: Python · Flask · PySpark · Databricks · Web Development

AI Backend Developer

Portal Telemedicina

• Integrated AI diagnostic outputs into clinical systems using Python, Django, Docker, and Kubernetes.
• Developed REST APIs for AI-generated medical imaging outputs and heatmaps.
• Built backend services supporting AI-assisted telemedicine workflows.
• Worked with GCP services including BigQuery, Data Fusion, Cloud Storage, and Looker.

🔹 Focus: Healthcare AI · Python · Django · REST APIs · Docker · Kubernetes · GCP

Full Stack Developer

Softplan

• Developed and maintained enterprise web applications using Python, Django, JavaScript, and PostgreSQL.
• Worked across the software development lifecycle, from requirements and implementation to deployment and maintenance.
• Built backend services and business workflows for legal technology applications.

🔹 Focus: Python · Django · PostgreSQL · JavaScript · Full Stack Development

AEC — Artificial Intelligence and Machine Learning

Collège LaSalle Montréal

Postgraduate Studies in IT Governance

Faculdades Integradas Promove de Brasília

Technical Degree in Data Processing

Instituto de Ciências Exatas