Staff / Senior Research Engineer - Algo Brain

Realize your potential by joining the leading performance-driven advertising company! As a Research Engineer on Brain, a global R&D team in Taboola's Algorithms Craft Center, you'll play a vital role in advancing the deep learning models behind Taboola's recommendations and ads. This role is based in our Taipei office. It is a chance to do research that ships: you'll turn new ideas from deep learning and recommender systems research into working models at Taboola's scale, rethink how they are built, and see your work directly improve a live production system. You'll also help shape the methodology algorithm engineers across the company use to develop and improve models. This role is open at Senior and Staff levels. At Senior level, you own model improvements end to end, from finding the opportunity to seeing the gain in production. At Staff level, you also set technical direction for a problem area, lead architecture changes across models and teams, and help the people around you grow through design reviews and mentoring. Both are individual contributor roles. To thrive in this role, you'll need: • Deep expertise in machine learning and deep learning, including designing model architectures for a specific problem domain, not only applying existing ones. Any large-scale domain counts: recommendation, ads, search, NLP, vision, speech. • Strong hands-on model development: you design, implement and train deep learning models yourself, in any framework (TensorFlow, PyTorch or others). • Proven depth, shown by at least one of: measurable improvements to machine learning models in production; papers at top venues such as KDD, RecSys, WWW, NeurIPS or ICML; top ranks in machine learning competitions such as Kaggle, KDD Cup or RecSys Challenge. • Strong engineering skills to build your models end to end, so that what works offline can run in production. • MSc or PhD in Computer Science, Mathematics, Statistics, Engineering or a related field, or equivalent research experience and expertise. • Clear written and spoken English, for working with teams across time zones. Bonus points if you have: • Experience with large-scale recommendation, ads ranking or CTR/CVR prediction models. • Experience with multi-task, multi-domain or user-sequence models. • Experience defining ML experimentation and optimization methodology: how model variants are evaluated, compared, tuned and diagnosed. How you'll make an impact: • Find where our production models for CTR, CVR and value prediction can improve fundamentally: architecture, representation, training objectives, training regime. • Design and implement new model architectures, adapting ideas from research and industry and creating new ones when existing methods do not fit. • Take proven improvements to production together with the teams that own each model, and make sure the accuracy gain seen offline holds in production. • Improve how models are developed across the company: standardize how models are evaluated, compared and optimized, and build tools that shorten the path from idea to a better model. Our tech stack: TensorFlow, Spark and Airflow, on a heavily customized internal platform. Why Taboola? If you ask Taboolars what they love about working here, they’ll tell you they’ve been empowered to realize their full potential while growing and learning from smart, talented peers. They’ll also share more about: • Adam Singolda, Taboola Founder and CEO, says: “ You can copy anything from another business, but you can’t copy a company’s culture .” • Well-being: Enjoy comprehensive benefits, a fully stocked kitchen, and location-specific perks. • Flexibility: Hybrid work model with 3 days in-office and the option to come in more often if desired. • Work with the biggest names: We