Barbara Macias
Zapopan, Jalisco
Summary
Senior Data Scientist with 6+ years of experience in natural language processing (NLP), statistical modeling, and deep learning (ML/DL). Proven ability to design scalable machine learning pipelines, optimize algorithms for real-world applications, and lead cross-functional teams to meet business-critical goals. Expertise in Python, R, TensorFlow, PyTorch, and AWS with strong problem-solving and communication skills. Passionate about delivering data-driven insights that drive innovation in healthcare and technology.
Overview
8
8
years of professional experience
1
1
Certification
Work History
Sr. Machine Learning Engineer
Sourcescrub
San Francisco, USA (Remote)
01.2022 - 09.2024
- Led communication and organization within the team and with other teams for 18 months after the team lead departed, resulting in improved performance and collaboration.
- Collaborated with multi-disciplinary product development teams to identify ML opportunities and integrate trained models.
- Prototyped machine learning applications and quickly determined application viability.
- Automated company classifications and description generation to scale from under 5 to 20+ million companies, significantly increasing deals by transitioning from manual to automated processes.
- Implemented an automatic re-training pipeline process collaborating with DevOps, an architect and humans-in-the-loop, setting up the model re-training, comparison, deployment, and performance monitoring, drastically reducing manual workload and including MLOps practices.
- Composed production-grade code to convert machine learning models into services and pipelines to be consumed at web-scale with optimal performance and load capabilities.
- Trained and mentored a new machine learning engineer, enhancing their independence and project delivery by conducting frequent peer reviews, one-on-one coaching, and providing infrastructure guidance.
- Supported AI-first transformation efforts with assistance to 8 engineers (3 lead engineers, 3 architects) and collaborating with DevOps to ensure accessible infrastructure for out-of-the-box model deployment and usage.
- Delivered state-of-the-art models tailored to company needs, including revenue estimation, valuation, and text quality assessment across diverse data sources like Glassdoor, LinkedIn, and human annotations.
Full Stack, Data Scientist & +
Personal Proyect - BiosignalsIT
Zapopan, Mexico
05.2021 - Current
- Developed a prototype in 4 months, integrating hardware selection, connectivity to a custom iOS/Android app, and AWS servers and databases for analyzing and storing biomedical signals and clinical data to assess risk factors.
- Consulted post-PRECISE project on the analysis of PPG signals in pregnant women with the aim of validate diagnostic capabilities in hypertensive cases against Doppler as the gold standard.
- Provided knowledge transfer post-PRECISE project supporting a researcher and undergrad student from the University of Strathclyde in the PPG signal analysis methods and applications.
Machine Learning Engineeer
Espressive
Zapopan, Mexico
02.2020 - 03.2021
- Developed in 3 months a new paraphrase model, the heart of the then current intent matching. The new model improved the most relevant pain points: 1.8x better with partial matches, 2x more consistent with different entity values, and 3x better at differing between positive and negative phrases
- Developed a text classifier, named entity recognition model, a model to generate phrases, and a couple of tools for data cleaning.
Algorithms Specialist
Soluciones Kenko
Zapopan, Mexico
04.2019 - 01.2020
- Developed a model in 3 months to approximate standard electrocardiograms (ECG) using data from the ECGlove. The ECGlove is a glove used by a doctor and must be positioned on the patient's chest.
- Validated the model in public databases prior to clinical trials
- Designed in 2 months the protocol for clinical trials according to international and national regulations
Research Trainee
BC Children's Hospital Research Institute
Vancouver, Canada
09.2016 - 12.2018
- Sponsored by Bill and Melinda Gates Foundation, CONACYT-COECYTJAL and MITACS
- Designed the clinical protocol for a PPG study in pregnant women, collaborating with a global health initiative (PRE-EMPT), team in Canada, England, Pakistan and Mozambique.
- Implemented an algorithm to automatically assess the quality of 17,000+ photoplethysmography (PPG) signals.
- Developed algorithms to extract PPG features from 11,000+ pregnant women and performed clinical data cleaning and selection.
- Increased understanding of gestational age and blood pressure effects on PPG signals.
- Identified a 3x higher risk of perinatal death in hipertensive pregnancies using the extracted PPG features.
- Developed strong working relationships with fellow researchers, fostering a positive and collaborative work environment.
Consultant
Pathonix
Vancouver, Canada (Remote)
11.2018 - 11.2019
- Evaluated client needs and expectations, defining clear goals for consulting engagements and maintaining strong relationships through regular progress updates.
- Performed research reviews and preprocessed data by cleaning and aligning PPG and blood pressure signals for analysis.
- Implemented an algorithm to estimate PPG signal quality and tested associations between blood pressure changes and PPG features in an animal model.
Education
Master of Science - Biomedical Engineering
University of British Columbia
Vancouver, Canada12-2018
Bachelor of Science - Biomedical Engineering
Universidad De Guadalajara
Guadalajara, Mexico08-2015
Skills
- Machine learning, model development & feature engineering: bayesian inference, probabilistic models, semi-supervised learning, anomaly detection, neural networks, ensemble methods, support vector machines, reinforcement learning, dimensionality reduction, transfer learning, random forests, clustering algorithms, k-nearest neighbors, time series analysis
- Soft skills: attention to detail, problem-solving abilities, teamwork and collaboration, organizational skills, data-driven decision making
- Natural Language Processing: Transformers, LangChain, GPT, OpenAI, Llama, LLM, HuggingFace
- ML Frameworks: PyTorch, TensorFlow, Trax, scikitlearn, onnx
- Data Analysis: pandas, numpy, spark
- Experiment tracking: MLFlow, Weights&Biases, TensorBoard
- Model optimization hyper parameter: Weights&Biases, RayTune
- Hosted notebooks management: Databricks, Google Colab, Amazon Cluster
- Model deployment and serving: Flask, Azure Machine Learning, RestAPI, httpx
- Model performance: locus, postman
- Data version control: Databricks FeatureStore & DeltaLake, DVC
- Development: Git, VisualStudio, agile methodologies
- Visualization: Altair, Seaborn, Matplotlib
- Data streaming: Kafka
- Model explainability: SHAP
- R & Matlab
- Databases: SQL, MariaDB
- App development: Kotlin, Xcode
Certification
Natural Language Processing Specialization - DeepLearning.AI - Coursera
Applied Data Science Specialization - IBM - Coursera
IBM Data Science Professional Certificate - IBM - Coursera
Languages
Spanish
Bilingual or Proficient (C2)
English
Advanced (C1)
Timeline
Sr. Machine Learning Engineer
Sourcescrub
01.2022 - 09.2024
Full Stack, Data Scientist & +
Personal Proyect - BiosignalsIT
05.2021 - Current
Machine Learning Engineeer
Espressive
02.2020 - 03.2021
Algorithms Specialist
Soluciones Kenko
04.2019 - 01.2020
Consultant
Pathonix
11.2018 - 11.2019
Research Trainee
BC Children's Hospital Research Institute
09.2016 - 12.2018
Master of Science - Biomedical Engineering
University of British Columbia
Bachelor of Science - Biomedical Engineering
Universidad De Guadalajara