What You Will Do:
- Architect Multi-Stage Recommendation Pipelines: Lead the design and implementation of our AI-Native Recommendation Pipelines, encompassing Generative Retrieval, Deep Ranking, and Re-ranking, serving millions of requests.
- Bridge Theory to Production: Act as the team’s primary "Paper2Code Ninja," distilling SOTA research (NeurIPS, KDD, RecSys) into robust, production-ready code optimized for sub-millisecond latency.
- Evolve Content Understanding: Direct the research on information-rich embeddings, utilizing unsupervised and generative approaches to create a unified representation of users and ads.
- Scale Innovative Tech: Own the transfer of innovative technologies into high-concurrency environments, ensuring large-scale generative models thrive under extreme production constraints.
- Act as a Team Multiplier: Lead architectural debates, mentor junior scientists, and foster a collaborative environment where constructive feedback pushes the team to excel.
Who You Are:
- The Scholar-Architect: PhD in CS, Stats, Math, or Physics (or MSc with extensive industry research) and a track record of publications in top-tier venues (SIGIR, KDD, NeurIPS, ICLR, etc.).
- RecSys Veteran: 4+ years of professional experience building large-scale recommenders in high-throughput industries (AdTech, E-com, or Social).
- Full-Stack Researcher: Deeply proficient in PyTorch with experience in high-concurrency optimization, model quantization, and distillation.
- Independent Driver: You don't need a roadmap; you create it. You thrive on solving "unsolved" problems with a direct impact on business ROI.
Fuel for the Journey: Benefits to Support Your Ambitions
- Invest in Your Future: Regular feedback and our development program support your growth, helping you expand your skillset and achieve your career goals.
- Easy Arrival to adjoe: From signing to settling in Berlin, we’ve got you covered. Need a visa? No problem. Ready to build your new life and career at adjoe in Berlin? We support every ambition—from learning German to a relocation bonus that helps you settle in and make Hamburg feel like home.
- Live Your Best Life, at Work and Beyond: We work in a hybrid setup with 3 core office days, plus flexible working hours. Enjoy 30 vacation days, 3 weeks of remote work per year, and free access to an in-house gym with lots of different fitness classes and mental health support through our Employee Assistance Program (EAP).
- Join the Community! Participate in regular team and company events, including hackathons and social gatherings. We work together, and we celebrate together, too.
Skills Required
- PhD in Computer Science, Statistics, Mathematics, or Physics (or MSc with extensive industry research)
- 4+ years of experience in building large-scale recommenders
- Proficient in PyTorch with experience in optimization and model distillation
What We Do
applike group is the leading company building in the app economy. Our technologies + businesses to cover the full value chain of the mobile apps market. Our companies work independently, but benefit from each others’ technologies and knowledge. We are 4 companies (JustDice, adjoe, Sunday, justtrack) under one roof of applike group. justDice: We build exciting app-discovery products for mobile gamers. With >50 million users we are the global market leader, while still growing rapidly every day adjoe: We build new technologies and ad formats that create a competitive edge for mobile app developers. Sunday: We create and publish extremely engaging and addictive mobile games with smart and player-friendly monetisation justtrack: One-stop software to automate your whole user acquisition process. This would be inconceivable without the 3 developer teams that form the core of applike group. We are 100 percent tech company where everybody is interested in computer science. We are always giving our best to get in touch with the newest technology in order to build the most relevant products. Our mission is to build software with the capability to distribute ads to every mobile device on earth.








