Machine Learning Engineer (Generative Models)

Updated 6 days ago

Job Description

This is an exciting role that offers a chance to work at the cutting edge of AI. In short, your main responsibility would be to harness the potential of recent breakthroughs in AI, especially in the area of generative models (text to image and large language models), to drive impact.


  • Understand recent advancements in machine learning, especially Generative AI and Large Language Models (LLMs), at a deep technical level. Use this knowledge to create novel technologies that were not possible before.
  • Keep up to date with developments in other areas like multi-modal models, computer vision, model distillation and fine-tuning. Some examples are trying out Segment Anything Model, Llama LLM, how to build something like LangChain.
  • Identify key problems and opportunities that can be addressed using these technologies.
  • Brainstorm and implement possible improvements to these technologies. Eg how to make a model smaller, optimize it for CoreML, and finetune for new data.
  • Stay up to date with the research landscape in these areas.


Technical (Required)

    • Well-versed with using at least one deep learning framework, preferably PyTorch.
    • Clear and thorough understanding of the original transformer architecture and its variants.
    • Strong understanding of how multi-modal and large language models work (eg data used for training, loss functions used, how to finetune etc).
    • Have a keen eye for detail - willingness to understand the nitty-gritty details, reading the source code when necessary.
    • Good foundational knowledge of probability and linear algebra.

Technical (Good to have)

    • Knowledge of HuggingFace API patterns (used across libraries like diffusers, transformers, accelerate etc).
    • Knowledge of docker and kubernetes.
    • Knowledge of diffusion models.
    • Publications relevant to ML.


    • Ability to think about potential applications and the impact of recent advances in AI (and not just focus on technical details).
    • Ability to communicate clearly and effectively.
    • Self-motivated and proactive: the ability to learn, figure things out and identify key problems to solve with little or no supervision.

Interview process

After you submit your application, the next steps are one or two video interview calls (with short written questions), followed by an onsite interview. Our interview process includes interesting exercises with our team members (e.g. pair programming, design brainstorming). Through this process, we hope you can get to know what it's like to work with us, and get to know our team and culture!

Please note that only applications through the link will be considered

No requirement for relevant working experience
1,700,000 ~ 3,000,000 TWD / year
Optional Remote Work
Personal Invitation Link
This is your personal referral link for job invitation. You'll receive an email notification when someone applied for the position via your job link.
Share this job
People who applied for this job also applied for
Logo of PicCollage 拼貼趣.

About us

嗨!我們是來自矽谷的新創團隊 PicCollage 拼貼趣,主要產品『PicCollage』是一個追求自由創造及分享的照片拼貼app,主要使用者來自美國、英國及日本等國家。目前全球下載量已突破 2.6 億,每個月 1500 萬穩定用戶量,並且持續增長中!

在PicCollage 拼貼趣團隊我們鼓勵自由創造,擁抱差異與多元文化,致力於打造『開放、學習、分享』的工作環境。另外,我們的新產品團隊近期以『影音編輯』作為開發重點,運用 Generative AI, Machine Learning、Data Analytics等技術,致力打造使用者友善,可以輕鬆上手的 app,再戰繼 PicCollage App 之後的下一個產品里程碑!

期待遇見更多 #builders #dreamers #thinkers 來激盪創意,加入 PicCollage 和我們一起做出富有創意、好玩的軟體產品到全世界 🌏



  • Our New Products
  • Life at PicCollage
  • Tech Blog
  • PicCollage: Growing a Culture of Learning, Exploration, and Trust
  • Always Be Learning! Workshop day at PicCollage ➤
  • PicCollage Office Tour ➤
  • 屬於 PicCollage 的黑客松 : InnoFest ➤


Mid-Senior level
1.7M ~ 3M TWD / year

Mid-Senior level
1.2M ~ 2M TWD / year

Mid-Senior level
1.6M ~ 2.5M TWD / year