As an ai engineer, you will be responsible for the development, implementation, and maintenance of sophisticated ai models, with a particular focus on diffusion, segmentation, and hybrid multi-modal (image + text) models built using pytorch. You will work closely with our data scientists and engineers to build scalable, high-performance solutions, leveraging the latest advancements in deep learning. The ideal candidate will have extensive experience in machine learning, specifically in neural networks, with strong skills in pytorch and a solid understanding of generative, image-based, and multi-modal models.
key responsibilities
* model development: design, develop, and optimize diffusion, segmentation, and hybrid multi-modal (image + text) ai models using pytorch for a range of applications, including image generation, image segmentation, text-based image captioning, and other advanced deep learning tasks.
* model training & evaluation: train and fine-tune large-scale neural networks and evaluate their performance using commonly used metrics and evaluation frameworks.
* research & innovation: stay up-to-date with the latest developments in ai, machine learning, and deep learning, particularly in diffusion models, segmentation, and hybrid multi-modal approaches. Propose and implement new techniques and algorithms.
* collaboration: work closely with cross-functional teams including data scientists, product managers, and software engineers to ensure the successful integration and deployment of ai models into production systems.
* optimization & scalability: improve the efficiency, scalability, and performance of models, optimizing them for both computational cost and accuracy.
* code quality: maintain high standards of coding, documentation, and version control. Write efficient, reusable, and maintainable code.
* troubleshooting & debugging: diagnose and resolve issues related to model performance, training stability, and deployment.
required qualifications
* experience: minimum of 3-5 years of experience in ai/ml development, with a focus on deep learning, diffusion models, segmentation models, or multi-modal models.
* technical skills: strong proficiency in python and pytorch for deep learning model development.
* experience with neural network architectures, including generative models (e.g., gans, vaes), segmentation models (e.g., u-net, mask r-cnn), and multi-modal models combining image and text data (e.g., clip, dall·e).
* familiarity with hybrid multi-modal model development, integrating both visual and textual data for improved performance.
* knowledge of computer vision techniques and related libraries (opencv, torchvision, etc.).
* familiarity with deep learning frameworks and libraries such as tensorflow, hugging face, and pytorch lightning.
* experience with python libraries for data manipulation (e.g., numpy, pandas), visualization (e.g., matplotlib, seaborn), and machine learning (e.g., scikit-learn).
* research & theory: strong understanding of the mathematical principles behind deep learning models, including optimization, loss functions, and model evaluation techniques.
* problem-solving: excellent problem-solving skills with the ability to apply cutting-edge techniques to real-world problems.
* collaboration: strong communication skills and the ability to work in a collaborative team environment.
* education: a bachelor’s or master’s degree in computer science, electrical engineering, mathematics, or a related field. A ph.d. is a plus.
preferred qualifications
* familiarity with cloud computing platforms (gcp is a must) for model training and deployment.
* familiarity with gpu cloud infra providers, such as baseten, replicate.
* experience with model deployment frameworks (tensorflow serving, torchserve, onnx, etc.).
* knowledge of distributed computing and parallelization techniques for large-scale model training.
* experience with version control systems (git, github, etc.) and agile development practices.
* contributed to published papers, open-source projects, or public-facing ai solutions.
as part of the evaluation process, you will be expected to use the following metrics and techniques to assess model performance:
* intersection over union (iou)
* pixel accuracy
* mean average precision (map)
* inception score (is)
for multi-modal models:
* text-to-image evaluation (e.g., clip score)
* bleu, rouge for text generation or captioning
* visual-semantic embedding evaluation
general metrics:
* precision, recall, f1 score
* area under the roc curve (auc-roc)
* loss functions (e.g., cross-entropy, mean squared error, hinge loss)
* training time and resource consumption
* model size (parameters) and inference time
* scalability and robustness on different hardware configurations
#j-18808-ljbffr