cv
Education, research, and professional experience. Download the full PDF CV above.
General Information
| Full Name | Ricardo Ignacio Pizarro Carreño |
| Location | Madrid, Spain |
| ricardo1459@gmail.com | |
| Languages | English (Full professional), Spanish (Native), Japanese (Intermediate), German (Beginner) |
Education
-
2023 - 2026 (Expected) PhD in Electronics, Advanced Electronic Systems (Intelligent Systems)
Universidad de Alcalá, Madrid, Spain - Thesis: Multi-view and landmark augmented deep learning model for HMI in autonomous vehicles.
-
2020 - 2021 MSc. Computer Science
Universidad Católica del Norte, Antofagasta, Chile - Thesis: Multimodal fusion technique for video sequences.
- Graduated with highest honors (Máxima distinción).
-
2014 - 2021 Engineer's Degree in Computer Science and Informatics
Universidad Católica del Norte, Antofagasta, Chile - Graduated with honors (Con distinción).
Research Experience
-
2023 - Present PhD Researcher
Universidad de Alcalá — Advisors: Dr. Luis M. Bergasa, Dr. Luis Baumela - Researching efficient video transformer architectures for human action recognition and temporal action detection in low-compute environments, combining token selection, native sparse computation, and custom GPU kernels.
- Designed native sparse 2D convolutions and custom CUDA kernels, released as an open-source PyTorch sparse-convolution library reaching up to 8x kernel speedup (SV-TAD, ECCV 2026).
- Benchmarked and validated PO-GUISE+ (IEEE T-ITS, 2026) for edge deployment on an NVIDIA Jetson Orin NX, reaching up to 57 FPS at FP16 with as little as 2.9GB of memory.
- Established SOTA results across multiple domains, including
- Driver Distraction (Drive&Act, 3MDAD, 100-Driver)
- Activities of Daily Living (Toyota Smarthome, NTU RGB+D 120)
- Temporal Action Localization (THUMOS14, ATTACH)
-
2022 - 2023 Multimodal Sharp Wave Ripple Detection
Donders Institute for Brain, Cognition and Behavior — Research Intern, Supervisor: Dr. Lisa Genzel - Ensured dataset integrity by implementing robust data labeling workflows and diagnostic tools to identify and correct annotation errors.
- Collaborated with a team of neuroscientists to perform feature engineering and analyze feature importance, directly improving classifier performance.
- Built a flexible deep learning framework using PyTorch Lightning and Weights & Biases to systematically benchmark CNN and Transformer architectures.
-
2020 - 2021 Multimodal Fusion Technique for Video Sequences
Universidad Católica del Norte — MSc. Thesis, Advisor: Dr. Juan Bekios-Calfa - Modified a multimodal fusion technique to leverage spatio-temporal relations in a video while using self-supervised learning.
- Achieved an improvement of 2.7 points over the baseline and obtained state-of-the-art results.
-
2020 Computer Vision Researcher — Image Segmentation in Mineral Flotation
Universidad Católica del Norte — Student Researcher, PI: Dr. Juan Bekios - Designed experiments to evaluate vision models (OpenCV, YOLO, ResNet-based architectures) for a novel computer vision application in mineral flotation devices.
- Developed a deployment-ready model for in-site testing and a performance tracking pipeline to verify its real-world performance.
-
2019 Reinforcement Learning Applied to Wave Energy
Universidad Católica del Norte — Student Researcher, PI: Roberto Cortés - Conducted experiments using reinforcement learning algorithms to control a wave energy converter in a wave simulator.
- Worked alongside a team of civil engineers to assess the performance of different algorithms compared to theoretical results.
-
2018 - 2019 Predictive Modeling Researcher — On-Time Graduation Projects
Universidad Católica del Norte — Student Researcher, PI: Brian Keith - Built a Random Forest model and ran statistical significance testing and feature-importance analysis to predict at-risk student project submissions.
- Published as Quelopana, Keith, & Pizarro (2024), Computer Applications in Engineering Education.
-
2016 - 2020 Teaching Assistant
Universidad Católica del Norte — Courses: Data Structures, Software Engineering, Artificial Intelligence, Intelligent Systems (M.Sc. level) - Designed and led weekly lectures for groups of ~40 students.
- Updated and translated course materials, including study problems, projects, and exams.
-
2020 Teaching Assistant
Universidad Adolfo Ibáñez - Planned deep learning projects related to image/video classification, NLP, and image segmentation.
Professional Experience
-
2019 - Present Founder & Chief Technology Officer
PIGNUS - Co-founded Pignus and led development of Rehaviour, a VR-based behavioral safety platform deployed with enterprise clients including Glencore, Copec, Equans, Albemarle, and Finning.
- Designed and trained the platform's AI/ML models, including unsupervised models trained and deployed on Azure Machine Learning to analyze in-VR behavioral data and identify safety-competency gaps.
- Implemented MLOps practices across the company's multi-cloud (AWS, Azure, GCP) infrastructure, covering the VR application backend, a web reporting platform (PHP/Laravel, MySQL), and data pipelines (Firestore, event-driven Cloud Run jobs).
- Led cross-functional development teams (VR, web, ML, infrastructure) and owned technical feasibility assessments and roadmap planning for AI and VR initiatives.
-
2021 Deep Learning Engineer
SERCOL - Benchmarked OCR and table-extraction approaches (Tesseract, TableNet, AWS Textract, Google Vision OCR) for document processing.
- Designed a scalable data pipeline to automate semi-structured document processing, cutting processing time by almost 75%.
Open Source Projects
-
2026 poguise
- Code and pretrained models for PO-GUISE+, a multi-task video transformer for efficient driver action recognition, including the Jetson edge-deployment benchmarking suite.
Technical Skills
-
Programming & GPU Systems
- Python, C++, CUDA — custom CUDA kernels and PyTorch C++/CUDA extensions for sparse and efficient computation
-
Deep Learning
- PyTorch, PyTorch Lightning, FlashAttention, scikit-learn, Weights & Biases
-
Computer Vision
- OpenCV, YOLO, ResNet, Vision Transformers (ViT, VideoMAEv2, InternVideo); object detection, classification, and segmentation
-
MLOps & Cloud
- Azure Machine Learning, AWS, GCP, CI/CD, model versioning & monitoring
-
Research Areas
- Vision Transformers; efficient & sparse computation (token pruning/merging, native sparse convolutions, adapter-based PEFT); temporal action detection/localization; multitask learning; human action recognition