Senior Data Scientist
Pendulum Intelligence, Inc.
Seattle, WA
Apply
JOB DETAILS
SALARY
$200,000–$200,000 Per Year
JOB TYPE
Full-time, Employee
SKILLS
Algorithms, Amazon Simple Storage Service (S3), Analysis Skills, Artificial Intelligence (AI), Benchmarking, C++ Programming Language, Comparative Analysis, Computer Engineering, Computer Science, Computer Vision, Cross-Functional, Data Science, Data Sets, Deep Learning, Documentation, Electrical Engineering, Establish Priorities, Image Processing, Machine Learning, Mentoring, Metrics, Model Validation, Performance Analysis, Performance Modeling, Prototyping, Python Programming/Scripting Language, Requirements Management, SQL (Structured Query Language), Test Plan/Schedule, Training Data Sets, Validation Testing
ADDITIONAL COMPENSATION
benefits include bonuses, health insurance, sick leave, paid time-off etc.
LOCATION
Seattle, WA
POSTED
21 days ago
1. Translate requirements into production-ready computer vision solutions by scoping requirements, building, comparing, evaluating and deploying state of the art machine learning/AI models, for tasks such as optical character recognition, hate symbol classification, guns detection, and more.
2. Prototype, engineer, benchmark, validate, and fine-tune deep learning models by writing productionquality code using Python, training and fine-tuning in TensorFlow and PyTorch, and rigorously assessing performance using mAP, IoU, precision–recall curves, and detailed error analysis.
3. Create, curate, analyze, and clean large datasets for labeling by defining schema and class distributions, writing comprehensive annotation instructions, training and guiding labelers, and conducting quality validation through spot checks, inter-annotator agreement analysis, and corrective feedback loops.
4. Run reproducible data-science experiments at scale with tools such as Amazon SageMaker, automate hyper-parameter sweeps, and track model metrics.
5. Apply advanced image-processing techniques (OpenCV, scikit, PIL) including contrast enhancement, sharpening, and super resolution to maximize downstream model accuracy.
6. Design and develop perception vision systems and computer vision systems, innovative methods for extracting keyframes from videos to analyze large volumes of video content efficiently.
7. Write labeling instructions and coordinate with data annotators to label curated data for training machine learning models.
8. Build and manage large, balanced datasets using SQL and Amazon S3; design annotation pipelines; and implement augmentation strategies.
9. Lead cross functional delivery: mentor engineers, coordinate with stakeholders, and turn experimental findings into roadmap decisions for the Pendulum platform.
10. Documenting experiments and findings by maintaining structured experiment logs and recording methodologies, hyperparameters, results visualizations, error analyses, and key insights to ensure reproducibility and facilitate knowledge sharing and identify challenges post data science experiments to help the engineering and product teams make decisions around prioritization, integration, and deployment of a solution.
11. Document workflows, experiments, and model performance via confluence for future reference and improvement.
May telecommute.
Requires a Bachelor’s (or foreign educ. equiv.) Degree in Computer Engineering, Electrical Engineering, Machine Learning, Computer Science, or a related field plus two (2) years’ experience in the job offered or related occupation.
Experience must include:
a. Designing and developing perception vision systems and computer vision systems.
b. Creating and curating datasets for labeling by defining schema and class distributions, writing comprehensive annotation instructions, training and guiding labelers, and conducting quality validation through spot checks, inter-annotator agreement analysis, and corrective feedback loops.
c. Engineering, benchmarking, and validating models by writing production-quality code using Python, and C++, training and fine-tuning in TensorFlow and PyTorch, and rigorously assessing performance using mAP, IoU, precision–recall curves, and detailed error analysis.
d. Documenting experiments and findings by maintaining structured experiment logs and recording methodologies, hyperparameters, results visualizations, error analyses, and key insights to ensure reproducibility and facilitate knowledge sharing.
e. Leading the perception development and testing efforts.
f. Building and managing large, balanced datasets using SQL and Amazon S3; designing annotation pipelines; and implementing augmentation strategies.
g. Using TensorFlow, PyTorch, Keras, OpenCV, scikitimage, ONNX, TensorRT, Docker, AWS (S3, EC2, SageMaker, EKS).
h. Algorithm & model development, comparison and evaluation: prototyping deeplearning architectures in PyTorch and TensorFlow.
i. Using hyperparameter tuning, distributed training, and logging with MLflow, ROC/AUC, and custom annotation tool (LabelImg).
About the Company
P