Deep Learning Engineer (f/m/d)

🙌 Who are we?
- A commercial open source company that focuses on MLOps platform for multimodal and neural search.- Founded in Feb. 2020, raised $37.5M in 20 months. Now a global team of 50 with four offices: Berlin (HQ), San Jose, Shenzhen, and Beijing.- One of the high-valued & high-potential AI startups in the world, featured on Forbes DACH AI30 2020, CBInsights AI 100 2021 & 2022.
😊 Benefits & Perks
💰 Competitive salary & stock options🌎 Multi-cultural & diverse team🎓 Numerous opportunities to present/attend top AI/OSS/industry conference🦄 Rapid career development opportunities alongside the company🏢 Central office in downtown Berlin, San Jose, Shenzhen, Beijing⛱️ Free snacks & drinks, monthly team events, flexible working hours, home office options💻 Macbooks & top-notch equipment
✨ Who do we want?
- You are passionate about multimodal intelligence and making it accessible to everyone.- You want to work with the latest technologies and are fascinated by AI/ML.- You are a fast learner and a team player and enjoy working in an async, distributed environment.- You are proactive and take ownership of your projects.- You have excellent communication skills in English.
💁 About this position
At Finetuning team, you will be primarily working on two things: 1. Research the state-of-the-art neural ranking algorithms. 2. Transform the algorithm into high-quality code inside Fientuner.  We expect you to have a background in deep learning and information retrieval. To be more specific: dense (text) retrieval, content-based image retrieval, multimedia information retrieval, video/music retrieval, deep metric learning, representation learning, and active learning. In short, we improve the quality of the embedding service to better matching users' information needs.

Responsibilities

• Research and follow up on the papers in the research domain. For instance, dense retrieval/image retrieval/short video retrieval, etc.
• Experimenting with the effectiveness of the research and prototype with vanilla PyTorch, evaluate results on search tasks/metrics.
• Integrate the prototype into the finetuner-core framework with high-quality code.
• Write a user-friendly interface (SDK) to allow users interaction with fine-tune tasks in the cloud and make sure the functionality is well documented.
• Research training/inference optimization strategies, e.g. distributed training, model pruning, and quantization.
• (Optional) We want to keep a presence in both academia and industry. You're welcome to prepare show & tell/talks and submit to workshops/conferences/meetups.

Qualifications:- You should have a Master's/PhD degree in Computer Science/Data Science/Artificial Intelligence/Machine Learning or related fields.- Your research/work/thesis/internship should be related to search or recommendation by leveraging machine learning techniques. - You understand the pros and cons of traditional search and neural search.- You either published papers or attend competitions and produce measurable results. You're able to follow and keep track of the latest development in the field.- You love Python and Pytorch. You can use the tools to bring research ideas into production-level code.- You are an easygoing person who would like to collaborate with team members and take the initiative to take on more responsibilities.- We are building an open-source community, and you would like to offer help to community members.

💼 Hiring Process
Candidates can expect the hiring process to follow the order below. Please keep in mind that candidates can be declined from the position at any stage of the process. 
- The first round is the CV screening, candidates will receive an email that contains a link for booking the next round. This process takes a maximum of one week.
- Qualified candidates will be invited to schedule a 30-minute screening call specifically on Zoom with one of our global recruiters. For engineering candidates, after this interview candidates will receive an email and be asked to complete an offline code challenge. On average the candidates can finish it in 30 minutes.
- Next, candidates will be invited to schedule Peer Interviews with team members from the relevant team. There are two rounds of Peer Interview, 1st is Technical Peer Interview and the 2nd is Team Peer Interview. For engineering candidates, the team will examine the quality of the offline challenge as well as you fundamental knowledge and coding skill during the Technical Peer Interview; one should also expect a live-coding challenge in 10 to 15 minutes. As long as candidates passed the Technical Peer Interview, they will be invited to talk with specific Team Lead in the Team Peer Interview stage. The interview will be more relevant to practical problem solving.
- Finally, candidates will be invited to schedule a 30-minute interview with CXO.
We will collect the feedback from all interviewers and make a decision in a maximum of two weeks (on average it takes 5 working days). Then the candidate will be invited to another 15-minute call with our recruiters to discuss the terms of the offer.