Design35 min
A language model in a loop that calls tools and acts on the results.Authentication, authorization and protecting data from abuse.Staying up and correct when parts of the system fail.Keeping data correct when many requests touch it at once.Cheap, durable storage for files and blobs, like S3.Core featuresExtreme durability (11 nines)Virtually unlimited scaleLow cost per GB, storage tiersPresigned URLs for direct uploadsWhole-object writes, no in-place editsWhen to useUser uploads, media and backupsLarge files rather than queryable recordsData lakes and long-term archives
PremiumRL Environment Platform
Run thousands of isolated, reproducible agent episodes for evaluation and RL training, and hand the trainer a reward it can trust.
What they’re really testing: Thousands of episodes must start from the same state in seconds, run whatever commands a model chooses without reaching the grader or the internet, and finish before the GPUs go idle waiting for the slowest one. An infrastructure failure scored as a zero becomes training signal. How do you make the reward trustworthy and the fleet fast at once?
Premium
This design is on the Premium plan
Readers with Lifetime Access read it in full. Already bought it? Sign in from the top bar.
What’s inside
- Primary concepts and the hard part
- Requirements
- API / Model
- High-level architecture
- Trade-offs and deep dives
- Possible follow-up questions