Machine Learning Engineer

About Us

We are a profitable and growing company, originating in Silicon Valley and now headquartered in Taiwan. We combine intuitive design with Creative AI tech to create inspiring products for millions of people worldwide.
We offer a fun, creative, and international workplace with competitive compensation, stock options, flexible hybrid work, free lunch, and more.
Ready to make a big impact with a talented team? Come create with us.

About The Role

As a machine learning engineer, you will build the machine learning models which power PicCollage’s creative and visual applications, using the latest in computer vision and generative AI technology. You will work with PMs, designers and other ML engineers to solve exciting problems and build new features which leverage the latest breakthroughs in AI. Specifically, you will work closely with engineers to train and accelerate models, and build novel solutions involving technologies like transformers and diffusion models.

What You'll Do

• Design and implement machine learning models for visually creative applications, including computer vision and generative AI.
• Keep up to date with the latest advancements in machine learning, and apply new technologies to improve our products.
• Collaborate with other engineers and cross-functional members to identify new challenges/opportunities and translate open problems into specific technical problems to work on.

Skills We're Looking for

• Experience in training or finetuning deep learning models, preferably in computer vision
• A deep and clear understanding of transformers, self- and cross-attention, ViTs.
• Strong grasp on a deep learning framework like PyTorch, Tensorflow or JAX (we use PyTorch here).
• A MS degree or higher in a related field (machine learning, computer science, etc), or equivalent experience.
• Ability to delve deep into topics and have a fundamental understanding of key concepts in ML. We prefer deep understanding of key concepts over having a shallow knowledge of a wide range of topics. Example of delving deep: you are working with CLIP model and want to visualize the attention maps for each transformer block to understand what’s going on.
• Good communication skills in English.
• Strong knowledge of linear algebra and probability.

Nice-To-Haves

• Experience in generative AI, particularly text to image models, is a great plus. Bonus points if you have experience training adapters (LoRA, ControlNet, IP-Adapter etc) for Stable Diffusion models.
• Has strong analytical skills
• Has has an artistic side
• Has experience in computer vision and/or generative AI.
• Published papers or research work related to ML.
• Familiarity with diffusers and diffusion models.

You should still consider applying even if you don’t have a lot of experience with diffusion models. A great candidate is someone who learns machine learning deeply, not only superficially. A candidate who developed deep expertise with any relevant ML topic will fit better than a candidate who only has limited understanding of diffusion.

There are three main stages in our interview process:

1. Screen calls (1-2 calls)
2. Take-home quiz & review
3. Onsite interview (3-4 hours visiting)
   

If we see a fit with the company, we will reach out to start getting to know each other. You can expect traditional discussions as well as participating in situational exercises. The goal of the interview process will be for us to see your skills and let you get to know the team and work culture to see if we are a match.