UCL Amazon Apprenticeship

Arctic Shores

패턴 그리기

아래처럼 비밀번호 맞추기

QP 위아래 둥근 짝수 홀수 빠르게 대답

QP 중앙에 좌우 화살표 방향 맞추기

직무 기술 테스트(예: 코딩, 알고리즘 문제 해결)가 아닌, 행동 및 성향 평가를 목적으로 하는 심리 테스트

Arctic Shores의 평가는 코딩 과제가 아닙니다.

이 평가는 심리학과 신경과학에 기반을 둔 게임 기반 평가로, 주로 개인의 성향, 사고 방식, 문제 해결 방식, 집중력, 패턴 인식 능력 등을 측정하려는 목적을 가지고 있습니다.

평가 내용은 다음과 같습니다:

게임 유형 과제

기억력 테스트

반응 속도 테스트

패턴 인식

숫자 문제 풀이

특정 역량 분석

시간 관리 능력

문제 해결 전략

포기 시점 판단 능력

Video Interview

웃으며 손짓하며 동작 발음 크게 크게 위아래 하고 말 너무 빠르게 하지말고 눈 깜빡이지 말고 마주치기

Tell us what attracted you to apply to Amazon, and why you believe an Apprenticeship is the right path for your career at this time?

Hello, the Amazon’s AI Apprentice program

impact in the industry than going for a PhD.

applied research AI control and ethical AI deployment

With my experience working in companies

I’d love to work on AI research that’s closely connected to

Hello, as a Master’s student, Amazon’s AI Apprenticeship is a perfect choice for additional research. I enjoy contributing to academia and am also excited about making an impact in the industry. First of all, the program’s focus on applied research fits perfectly with my interests in AI control and ethical AI deployment. Through my experience working in companies such as Kakao Mobility and building products at hackathons, I realized that I enjoy contributing to real-world applications while communicating with engineers and customers. Based on this, I believe that my research knowledge in Interpretable AI, specifically Mechanistic Interpretability led by Anthropic, will allow me to contribute to Amazon during the AI Apprentice program.

1 minute for 8 questions and 3 times for recording

Multiple Tabs: Please close all other tabs and applications, including any VPN you might be using or you will not be able to complete your interview.

Once you start the interview you may also have the option to re-record your answers at any time by using the Re-record button. Note: this is not available for all interviews or questions.

Tell us why you want to work for us (3:00)

because it focuses on applied research which aligns with my interests in AI control and ethical AI deployment
During my first term of MSc at UCL, sparse autoencoders, Hackathon
At Kakao Mobility, for efficient data generation for autonomous driving
I actively follow cutting-edge developments in inter/pretable AI with my own ideas
I’m eager to build on my experiences pushing the boundaries of AI innovation.

Hello, my name is Seonglae Cho, and I am a Master's student at UCL, majoring in AI for Sustainable Development. I want to join Amazon’s AI Apprentice program because it focuses on applied research, which aligns with my interests in AI safety and control. The opportunity to develop and test AI models while working on ethical AI deployment is exactly what I’m looking for to show up my research ideas.

During my first term of MSc at UCL, I worked on projects applying sparse autoencoders to interpret large language models, focusing on extracting features from large language models to control their behavior. This research efforts shows my ability to design innovative solutions and led to winning the Holistic AI Research Hackathon.

At Kakao Mobility, for efficient data generation for autonomous driving, I refactored a critical 3D mapping module from C++ to Rust, improving its cross-platform compatibility and reliability. This showed my ability to handle low-level systems and implement efficient algorithms that improves high level application performance.

I actively follow cutting-edge developments in interpretable AI, including Sparse Autoencoders, weight and gradient interpretability, and steering vector with my own ideas. This helps me stay propose several problem statements and suggesting innovative methods.

I’m eager to build on my experiences and contribute to Amazon’s work in pushing the boundaries of AI innovation.

Tell us about a specific time where ;you went above and beyond (1:00)

One time, a teammate accidentally deleted an entire Kubernetes namespace
Another example is during the RTSum research project
Lastly, while working on research project ReSRer,

One time, a teammate accidentally deleted an entire Kubernetes namespace during deployment, which caused critical downtime. I stayed up all night to restore the system, and afterward, I implemented fine-grained access controls to prevent similar incidents.

Another example is during the RTSum research project, where the semantic extraction server was a major bottleneck. I solved this by scaling Docker instances and introducing a reverse proxy, improving processing speed by 4x.

Lastly, while working on ReSRer, I initiated the project with an idea based on information density. When our team faced the limitation of performance, I used confusion matrix metrics to analyze the issues and collaborated with my teammates to find a better direction. This shows not only my problem-solving skills but also my ability to contribute effectively to research discussions.

Leadership Principles

What are your greatest strengths?

Name two of Amazon's Leadership Principles that resonate with you the most Leadership Principles. And for each one, provide a specific example from your life, education, or work experience that demonstrates how you've portrayed this Leadership Principle?

The first principle is Ownership since I always take responsibility for my work. For example, When I left my previous company, Kakao Mobility, I handed over everything to my successor and wrote tons of documents even up to the last minute. My goal was to keep the project alive ,because I had a strong sense of attachment for the project. And recently, I heard, they are still using my team’s solution.

The second principle is Learn and Be Curious. While I’m focused on my current projects, I operate in two modes: project mode and curiosity mode. After work, my curiosity drives me to explore new ideas. For instance, I actively follow cutting-edge interpretable AI, including Sparse Autoencoders for my own ideas. This habit fuels my endless personal development.

What motivates you to pursue an apprenticeship specifically in Artificial Intelligence? Describe three transferrable skills you posess that you believe are crucial for success in this role and why?

The first crucial skill is engineering skill for optimization. Researchers and scientists sometimes overlook computational costs; however, cost-effective computing is necessary for enterprise-level solution. I possess experience with Rust and C++ and parallel training with understanding MPI. The second is research ability. I have published a paper and also actively participating in the Interpretability community with my own ideas. Finally, productivity matters. I’ve developed applications in a company and published AI services like MBTI GPT. Through that experience, I am aware of the importance of fast iteration and the ability to switch between formality and informality to manage my productivity.

What are your greatest weaknesses?

I can be a bit rebellious at times, which makes me overlook the value of simple, conventional approaches. I’m also a bit of a perfectionist, so I sometimes spend too much time getting caught up in small details when working on research.

What programming languages have you used?

python, typescript, rust, c++, AHK

Describe the process you use for writing a piece of code, from requirements to delivery.

Ability to code-switch between formalism and informality is important

How do you design scalable applications? Walk us through your process.

Frugality - Developers and scientists can sometimes become too focused on logic and overlook computational efficiency or cost. However, optimized algorithms and cost-effective computing are tightly interconnected with real-world service deployment. I am both a researcher and a cost-aware engineer in enterprise-level software.

What would you do if a coworker asked you to review their code, and it was full of errors?

관계를 망치는 피드백은 비효율적. 상대에 도움이 되려고 하고, 다른 의도없다고 알리며 질문으로 시작하는 피드백

How do you feel about a collaborative work environment?

Company experiences, labs, and hackathons have given me the confidence to work with people from diverse backgrounds on a variety of projects. I’ve taken on roles as both a leader and a member, which has helped me adapt to different team dynamics.

Indeed, working efficiently through communication with a team is always a challenge, but it’s one I embrace. I feel motivated about collaborating with such competitive colleagues at Amazon, and I’m excited about the opportunity to grow and learn together.

Bias for Action - When I receive important feedback or hear about an urgent issue from customers, I act immediately, bypassing complex priority systems. I focus on testing, verifying, identifying, and addressing the underlying root causes of problems with urgency and flexibility, avoiding unnecessary delays or meaningless inaction.

Customer Obsession - I always ask for feedback from internal customers and stakeholders to improve my product. This obsession has always been the foundation of successful project development.

Invent and Simplify - My projects always aim for simplicity. Even when complex features were requested, I worked to simplify and abstract them to make the functionality intuitive.

Insist on the Highest Standards - When there are multiple ways to implement a solution, I always choose the best one, even if it requires more effort. From experience, I know that initial implementations often last a long time with maintenance. My goal is to create responsible, long-lasting software that meets my high standards.

Deliver Results - Ultimately, delivering results is the most important aspect in the company. Every decision I make and my priority system revolve around achieving outcomes. I pride myself on always making amazing results.

Interview in 5 Easy Steps

We want you to have the most enjoyable and relaxing interview experience - please read these tips before you get started to put your mind at ease: 1 Access t

https://support.willo.video/article/64-5-easy-steps-to-your-interview

Leadership Principles

We use our Leadership Principles every day, whether we’re discussing ideas for new projects or deciding on the best way to solve a problem. It’s just one of the things that makes Amazon peculiar.

https://www.amazon.jobs/content/en/our-workplace/leadership-principles

{ai: "b4e3f29d-2428-4ad6-8b85-9febd79b8a0a", p: "AP-KVTZIG94IZDZ-2"}

ai: "b4e3f29d-2428-4ad6-8b85-9febd79b8a0a"
p: "AP-KVTZIG94IZDZ-2"

{"ai":"b4e3f29d-2428-4ad6-8b85-9febd79b8a0a","p":"AP-KVTZIG94IZDZ-2"}

Many of the questions felt as though they were particularly tailored to filter out those with autism/ADHD/Dyslexia/etc. So it wouldn't surprise me if this one's the same

Arctic Shores assessments - what the hell was that

10 votes, 21 comments. I just completed an Arctic Shores assessment, and fair to say, there were some good parts and bad... so I thought. I got the…