As a Data Scientist and Machine Learning Engineer, I specialize in developing cutting-edge AI solutions. My expertise spans forecasting, classification, and regression problems, utilizing both traditional ML and deep learning techniques to deliver impactful results.
Beyond technical skills, I excel in leading and coordinating data science teams. I provide comprehensive mentorship in technical aspects, business strategy, and project execution, ensuring the successful delivery of AI solutions that align with business objectives and drive growth.
Work Experience
Machine Learning Engineer
Crossbridge Global Partners Jobs & Career
October 2024 - Present
Delivering machine learning and AI solutions for drug discovery, drug design, and drug optimization in the pharmaceutical sector.
Projects:
1. Representation learning:
- Trained foundational models from scratch to generate robust data embeddings for downstream tasks
2. Foundational models for drug discovery and design:
- Trained foundational models from scratch to generate robust data embeddings
- Adapted foundational models to predict molecular properties across diverse pharmaceutical scenarios
- Implemented custom loss functions for molecular property prediction
3. Sales forecasting and inventory optimization:
- Developed a framework to run forecasting models with different model flavors including statistical models, traditional ML, and foundational models
- Implemented few-shot forecasting using both traditional ML and foundational models
- Reduced global supply-chain errors by 60% using few-shot learning techniques
- Implemented automated model selection based on performance metrics
Tools & Technologies:
- Languages: Python
- Frameworks & Libraries: Pytorch, Numpy, Pandas, HuggingFace Accelerate
- Cloud & DevOps: AWS (S3, Batch, CodeCommit, CodeBuild, CodePipeline, Step Functions)
- MLOps: Docker, Terraform
Data Scientist
Banco Pichincha
January 2024 - October 2024
Developed machine learning models for various banking products and solutions in the Center of Excellence (CoE) of Data and Advanced Analytics.
Projects:
1. Data labeling and cleaning:
- Developed RAG systems with vectorial databases and large language models to clean and classify messy data
2. MacVoc:
- Developed a deep learning model using transformer architecture to automatically classify complaints from various bank communication channels
3. Client Acquisition:
- Created an analytical solution to identify users most likely to need the remittance product
- Resulted in a 200% increase in marketing campaign yield
4. People Churning:
- Developed an ML model to predict customer churn probability for future months
5. Next Product to Buy:
- Designed a recommender system for all bank clients using the TF-IDF technique from Natural Language Processing (NLP)
Tools & Technologies:
- Languages: Python
- Frameworks & Libraries: Numpy, Pandas, PySpark, Matplotlib, Seaborn, Scikit-learn, TensorFlow, Keras, PyTorch, HuggingFace
- Cloud & DevOps: AWS (S3, Bedrock, Sagemaker), Azure (Databricks, DevOps)
- Other: Stratio (Rocket, Intelligence, Discovery)
Data Scientist Junior
Bi Solutions S.A
December 2022 - January 2024
Worked on projects involving data science, machine learning, and machine learning operations (MLOps) in the cloud.
Projects:
1. Airport Store Sales Prediction:
- Forecasted sales for airport stores and optimized flight-gate assignments
- Increased sales by 30%
2. Loan Default Prediction:
- Developed machine learning models in a request-response architecture
- Reduced new model implementation time by 150%
- Achieved a 3-second response time for 252 stores in Ecuador
3. Image Recommendation System:
- Created a recommendation engine using Encoders and Vision-Transformers (Deep Learning) to suggest similar tiles based on images
- Boosted cross-selling by 20%
4. Item Forecasting and Inventory Optimization:
- Used machine learning models to forecast item consumption, avoiding overstocking or understocking
- Estimated needs for new items (cold-start forecasting)
5. Car Fault Prediction:
- Calculated the probability of failure for various car components and their occurrence mileage
- Reduced leasing costs by 35%
6. Customer Segmentation:
- Identified customer clusters using RFM analysis and machine learning techniques
- Uncovered distinct patterns and behaviors in purchasing
Tools & Technologies:
- Languages: Python
- Frameworks & Libraries: Numpy, Pandas, Matplotlib, Seaborn, Scikit-learn, TensorFlow, Keras, PyTorch, Streamlit, FastAPI
- Cloud & DevOps: AWS (Sagemaker, ECR, App Runner, S3, EventBridge, Batch), Azure (Databricks, App Runner, Container Registry), GCP (Cloud Run, Artifact Registry, Dataproc)
- MLOps: Docker, Git, GitHub, DagsHub, MLFlow
- Other: KNIME, Alteryx