Sebastian Cepeda

Senior ML Engineer with an MSc in Electrical Engineering and 10+ years building and deploying production ML systems. Specialized in deep learning, computer vision, LLMs, and generative AI across healthcare, finance, retail, and security domains. Proven track record of shipping models that improve detection accuracy, reduce churn, and automate workflows at scale.

Experienced across the full ML lifecycle: data pipelines, model training, MLOps, and cloud deployment.

Sr. Machine Learning Engineer at Tecla / Incyte (2024-10 – 2026-03)

Researching and developing generative AI models and LLM-powered solutions for clinical research and medical automation.

• Researching and developing generative AI models (diffusion) for synthetic tabular data in a clinical research context
• Built an LLM-powered embedding pipeline over medical literature search results; applied clustering to surface actionable insights and statistics
• Designed prompt engineering workflows with Claude (Anthropic) to automate generation of clinical study documentation, significantly reducing manual effort

Machine Learning Engineer at Intuition Machines (2021-07 – 2024-07)

Built end-to-end MLOps pipelines and developed CNN models for threat detection with production deployment and monitoring.

• Built an end-to-end MLOps pipeline for training and deploying models on 2M+ data points, including CI/CD, automated testing, and model versioning
• Developed a CNN model for motion-based threat detection, improving malicious activity detection by 20%; optimized for real-time inference using ONNX and C runtimes
• Improved production customer-specific models, boosting recall from 0.50 to 0.78 with no increase in false positive rate
• Implemented SHAP-based model explainability and automated alerting for data drift and performance degradation in production

Senior Data Scientist at Consorcio (2020-03 – 2021-07)

Developed and deployed churn prediction models, recommender systems, and NLP pipelines for customer relationship analysis.

• Developed and deployed a churn prediction model trained on 15M data points, integrated with Salesforce to power marketing and retention campaigns
• Built and deployed a recommender system that automated marketing campaign execution, reducing manual segmentation effort
• Implemented NLP and sentiment analysis pipeline (text + speech-to-text) for customer relationship analysis

Data Scientist at Cencosud (2018-12 – 2020-02)

Developed production sales forecasting models and real-time pedestrian detection systems for retail logistics.

• Developed production sales forecasting models for retail logistics across millions of SKUs for paris.cl and Colombian supermarkets; scaled data processing with Dask and PySpark
• Built a real-time pedestrian detection and tracking system (CNN + Kalman filter) to monitor customer foot traffic across stores

Software Engineer – Computer Vision at WoodtechMS (2015-04 – 2018-12)

Developed embedded computer vision software for industrial material measurement and access control products.

• Developed embedded computer vision software for industrial material measurement, deployed to customers globally
• Owned the full software lifecycle of an industrial access control product, from design to production release
• Mentored junior engineers and interns on signal processing, Kalman filters, and regression models

MSc in Electrical Engineering – Universidad de Chile (2014 – 2016)

Minor in Computer Science – Universidad de Chile (2011 – 2013)

BSc in Electrical Engineering – Universidad de Chile (2008 – 2016)