Cristian P
Verified Expert in Engineering
Deep Learning Developer
Cristian has four years of experience in Computer Vision (CV) and Machine Learning (ML). He has worked in natural language processing, autonomous driving at startups, and face tracking at Oculus. Running throughout the whole ML system lifecycle, from training models to testing and serving behind an application interface. Cristian enjoys curating datasets and can communicate with all teams to articulate results while researching tooling, frameworks, and papers needed for an optimal solution.
Portfolio
Experience
Availability
Preferred Environment
PyTorch, Python 3, JavaScript, Bash, NumPy, Pandas, SQL, C++11
The most amazing...
...project I've built is a data pipeline for face key-point estimation, an evaluation of object classification, and multi-class, multi-label classification.
Work Experience
Research Engineer
Meta
- Managed language models pre-training and fine-tuning to detect privacy-sensitive artifacts, trained and optimized the BERT model, which improved evaluation metrics by 2% over the baseline.
- Collaborated with XFN teams and sourced 3–5 million data points to create training datasets and ingest them into BERT models.
- Facilitated implementation of service API for serving the BERT model unlocking new use cases for three teams.
Research Engineer
Oculus
- Designed and implemented a resilient data pipeline for ingesting, prioritizing, and sampling video data for key-point labeling to automate acquiring raw data and getting it labeled.
- Analyzed and benchmarked different models for face-tracking and collaborated with the mobile device team to deploy the best face-tracking model to specialized hardware.
- Collaborated with the data operations team and implemented key-frame detection and active learning sampling methods, reducing the need for data annotation by 30%.
Computer Vision Research Engineer
Cognition Factory
- Implemented a modular and configurable pipeline to collect labeled images on new objects and fine-tune object detection models.
- Investigated and re-ordered Numpy operations to improve the latency of Yolo darknet Python implementation.
- Developed a background segmentation feature to crop out objects so that the object classification models can be trained on data generated with that object.
Computer Vision Intern
Motional
- Developed a plane detection and segmentation algorithm to partition lidar maps.
- Created a feature in the labeling tool to detect inconsistent lidar maps to reduce annotation time by 20%.
- Implemented UI features in the labeling tool using the Qt framework.
Part-time Computer Vision Engineer
Cognition Factory
- Implemented a fish-eye lens camera calibration method from a paper.
- Developed triangulation methods and integrated them into a SLAM pipeline.
- Created an end-to-end monocular up-to-scale visual odometry pipeline.
Experience
Object Multi-Label Classification
Education
Master's Degree in Machine Learning and Computer Vision
Technical University of Munich - Munich
Bachelor's Degree in Computer Science
Technical University of Munich - Munich
Certifications
Structure Machine Learning Projects
Coursera
Neural Networks and Deep Learning
Coursera
Skills
Libraries/APIs
PyTorch, NumPy, OpenCV, Pandas, PCL, Matplotlib
Languages
Python 3, Python, SQL, JavaScript, Bash, C++11, C++, PHP
Other
Deep Learning, Object Detection, Object Classification, Image Processing, Ceres, Qt 4, Machine Learning, Natural Language Processing (NLP), Language Models, GPT, Generative Pre-trained Transformers (GPT)
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring