The Reality of Working with GenAI at Amazon: Opportunities and Challenges
In recent years, the buzz surrounding generative AI (genAI) and large language models (LLMs) has been heating up, particularly at tech giants like Amazon. It almost feels like any internal project that doesn’t align with these trending terms is at risk of falling flat. As a former employee in Amazon’s genAI direction, I want to shed some light for aspiring professionals at Amazon aiming to step into the realms of LLM and genAI. Today, let’s delve into one of the notable areas to be cautious about: AWS Bedrock.
What is AWS Bedrock?
During my time at Amazon, I was part of a core group within Bedrock, a service that provides access to various foundation models, including those from Anthropic. We proudly hosted a significant portion of Bedrock’s traffic, and with every project launch, our CEO, Andy Jassy, would take to social media to showcase our work.
Pros of Working in AWS Bedrock
- Access to Cutting-Edge Models: One of the biggest advantages is the opportunity to engage with state-of-the-art models from Anthropic. If you’re lucky, you may even get involved in projects in collaboration with them.
- Valuable Kubernetes Skills: You’ll gain extensive knowledge about Kubernetes (k8s). Regardless of which model ultimately succeeds in the AI race, it’s likely that top performers will be hosted on k8s. Mastering this can equip you to adapt to future changes effectively.
- Supportive Team Environment: The camaraderie among team members is commendable. In tough times, humans instinctively band together, and a supportive environment can make a significant difference.
- High Impact Projects: The projects you’ll work on have substantial impacts, starting from Level 7 and 8 and frequently reporting to L10 and L11 positions. Some projects may even reach the level of being presented directly to Andy Jassy.
Cons of Working in AWS Bedrock
- Monotonous Tasks: Besides a few interesting initiatives, many tasks feel repetitive and akin to being a “human autoscaler.” This can feel like a waste of your potential.
- Challenging On-Call Experience: The infrastructure launches are often rushed and poorly constructed, making being on-call a dreadful experience. Can you imagine handling a week with 108 Sev 2 incidents? If you’re not resilient, best to avoid this path!
- Leadership Challenges: Often dominated by leadership who might primarily blame subordinates rather than taking responsibility. This toxic environment can create a challenging workplace culture.
- A Limited Credit System: While the projects have high visibility, recognition often goes to higher-ups, leaving little for the hard-working team members. Promotions can be exceptionally difficult, with last year’s promotions primarily favoring select individuals.
Final Thoughts
With the pros and cons outlined, I would rate my experience in AWS Bedrock as a 3 out of 5 stars 🌟. If you’re cognitive of these challenges and can navigate the landscape wisely, it could still be an excellent learning experience.
What’s Next?
In my next blog, I will dive into the world of Alexa and what it has to offer for further professional growth. Stay tuned!
#WorkLife #GenAI #Amazon #AI