Enrique G. Ortiz

Senior Applied Scientist and computer vision lead with 10+ years architecting high-performance Edge AI and recognition systems. Expert in data-centric ML, using synthetic data pipelines and GenAI to cut model development cycles by 70%. Proven track record deploying optimized, real-time inference on constrained hardware. Deep expertise in object detection, temporal modeling, and scaling ML infrastructure 0-to-1.

Experience
April 2023 — January 2026

Axon Enterprise, Inc.

Senior Research Scientist II
  • Architected a synthetic-data pipeline using diffusion-based image synthesis and VLM-guided auto-labeling for redaction detectors — cutting research-to-production from 8 months to 2–3, and supplying 30% of all training data.
  • Led deployment of advanced redaction models, improving precision/recall by 5–10% and capturing 40% of customer redaction usage.
  • Co-developed an org-wide standard for multi-object tracking and detection, adopted across 3 research and product teams to unify training and evaluation.
  • Tech-led a 3-person team on a confidential pre-launch computer vision initiative, owning technical direction and architecture in an ambiguous 0-to-1 problem space.
  • Delivered optimized face/head detectors achieving a 32% recall improvement at 90% precision.
March 2021 — October 2022

Peloton Interactive, Inc.

Staff Machine Learning Engineer
  • Sole owner of ML algorithms for temporal exercise recognition and repetition counting, achieving a <2% off-by-one error rate.
  • Built an automated evaluation pipeline that moved model releases from monthly/bimonthly to weekly.
January 2020 — February 2021

Samsung Next

Senior Staff Machine Learning Engineer
  • Engineered a real-time, on-device pose-based action recognition system for iOS, achieving 98% accuracy in exercise recognition.
  • Hired and led a team of 3, establishing Samsung Next's initial ML capability alongside a scalable pipeline with lineage, versioning, and experiment tracking.
June 2014 — January 2020

Sighthound, Inc.

Lead Computer Vision Scientist
  • Engineered a multi-task convolutional network for face analytics in production identity systems, achieving <1% false accept/reject rates in commercial deployments.
  • Enhanced License Plate Recognition accuracy by 20% through architecture adjustments and data strategy.
Technical skills
Languages
CorePython · C++
Deep learning
CorePyTorch · CNNs · Transformers · Computer vision · Object detection · Spatial-temporal modeling · Face recognition · Human action recognition · Hugging Face (Transformers · Accelerate · Datasets)WorkingTensorFlow · PyTorch Lightning · Optuna
Generative & synthetic data
CoreDiffusers · Stable Diffusion · Object-insertion augmentation · VLM-guided auto-labeling (GPT-4V · Gemini · Claude)
Vision-language models
CoreCLIP · SAM · Grounding DINOWorkingSigLIP · Florence
Edge & inference
CoreTensorRT · ONNX / ONNX Runtime · Triton Server · CUDA · Quantization · Real-time optimization (iOS / Edge)WorkingCore ML · TFLite · Qualcomm SNPE · OpenVINO · TorchServe · FastAPI
MLOps & tracking
CoreHydra · Weights & Biases · MLflow · Automated evaluation pipelines · End-to-end orchestration · Experiment & dataset tracking · High-throughput inferenceWorkingTensorBoard · DVC
Data & annotation
CoreAlbumentations · Dataset versioning & lineageWorkingLabel Studio · CVAT · Roboflow
Infrastructure & CI
CoreDocker · GitHub ActionsWorkingAWS (S3 · EC2 · SageMaker) · GCP (Vertex AI · GCS) · Azure ML · Jenkins · pre-commit · pytest · Ruff · mypy
Classic CV & SciPy stack
CoreOpenCV · NumPyWorkingSciPy · scikit-learn · Pandas · supervision
Publications
CVPR 2014

A. Dehghan, E. G. Ortiz, R. Villegas, M. Shah. “Who Do I Look Like? Determining Parent-Offspring Resemblance via Gated Autoencoders.” IEEE Conference on Computer Vision and Pattern Recognition.

CVPR 2013

E. G. Ortiz, A. Wright, M. Shah. “Face Recognition in Movie Trailers via Mean Sequence Sparse Representation-based Classification.” IEEE Conference on Computer Vision and Pattern Recognition.

CVIU 2013

E. G. Ortiz, B. C. Becker. “Face Recognition for Web-Scale Datasets.” Elsevier Computer Vision and Image Understanding.

CVPR-W 2013

B. C. Becker, E. G. Ortiz. “Evaluating Open-Universe Face Identification on the Web.” IEEE CVPR Workshop on Analysis and Modeling of Faces and Gestures.

IEEE FG 2008

B. C. Becker, E. G. Ortiz. “Evaluation of Face Recognition Techniques for Application to Facebook.” IEEE International Conference on Automatic Face and Gesture Recognition.

Thesis 2014

E. G. Ortiz. Taming Wild Faces: Large-Scale, Real-World Face Recognition in Still and Video Imagery. Ph.D. dissertation, University of Central Florida.

Patents
US 9,710,716

“Data-Enhanced Video Viewing System.” Sighthound, issued 2018.

US 9,336,433 B1

Unconstrained video face recognition, issued from the CVPR 2013 work. University of Central Florida.

PCT/US22/22879

“Connected Fitness Systems and Methods.” Peloton Interactive, filed 31 March 2022. Application pending.

WO2024/064703

“Repetition Counting within Connected Fitness Systems.” Peloton Interactive, filed 19 September 2023. US applications pending.

Education
UCF

Ph.D., M.S., B.S. — Computer Engineering

University of Central Florida · Center for Research in Computer Vision
  • Dissertation: Taming Wild Faces: Large-Scale, Real-World Face Recognition in Still and Video Imagery
  • Advisor: Prof. Mubarak Shah, Director, Center for Research in Computer Vision