Summary
What you’ll impact
The role is for a senior-level Machine Learning Engineer or Research Scientist to design, build, and scale feed ranking and personalization systems at the organization. The individual will create and deploy advanced ML models, including LLMs and multimodal networks, and collaborate closely with engineering and product teams to improve user engagement and retention.
Responsibilities
What you'll do
- Architect, train, and deploy our 0-1 core machine learning models (e.g., Transformers, Multi-modal Networks, Graph Neural Networks, Large Language Models) to generate rich embeddings for users, creators, and diverse content.
- Build content understanding and safety models across text, images, video, and audio: classification and tagging, quality and originality signals, and the multimodal representations that retrieval runs on.
- Partner closely with the Feed Ranking and Backend engineering teams to ensure your models and representations efficiently power downstream retrieval and personalization systems.
- Fine tune & post train LLMs, owning training pipelines and eval flows.
- Design rigorous offline evaluation frameworks to measure representation quality, and collaborate on A/B tests to tie underlying model improvements directly to engagement and retention metrics.
- Stay up-to-date with the latest research in deep learning, foundation models, and multi-modal AI, bringing innovative architectural ideas and industry best practices to the team.
Requirements
What you’ll bring
- Bachelor’s or higher degree in computer science, machine learning, statistics, math, or a similarly quantitative field.
- 5+ years of industry experience as a machine learning engineer or research scientist focused on deep learning and model architecture (2+ years if coming from a relevant PhD), having trained and shipped multiple models to production.
- Proven hands-on experience building, training, and deploying Deep Learning, NLP, Computer Vision, or Representation Learning models from scratch. Experiences working at a social discovery or content-heavy platform is a plus.
- Deep proficiency in Python & PyTorch, fine-tuning techniques, and multimodal models
- Experience with large-scale data processing technologies
- A strong understanding of embedding spaces, representation learning, and how they apply to consumer social dynamics and UGC is preferred.
- Thrive in ambiguous, fast-paced environments, enjoy building end-to-end systems, and are biased towards action and product-attuned building.