awsanswers

Inspect headers for the request you need to filter, serve the 403 from the load balancing tier

I was fascinated by this as a young teenager. I still love it

If you're in tune with animals and spend time around a parrot, it's obvious there is a lot going on in their minds. They have incredible memories and their own understanding of their world. It looks simple to us but they are not simple creatures. That being said, I don't know how a bird lover can keep a bird in a cage.

It's predictable so you run defense around it with prompting, validation and model tuning. It generates volumes of working code in seconds from natural language prompts so it's extremely business efficient. We're talking about tools that generate correct code to 95% of a solution, the follow up human and automated test review, and second coding pass to fix the 5% are a non issue.

I had ChatGPT explain why this is works as a unlimited hotpot. Very cool: "It works because carriers detect tethering partly by looking at TTL values. Android subtracts 1 from TTL when hotspot/tethering packets are forwarded, so the carrier sees a different TTL and knows the traffic came from a tethered device. Your eBPF program increments the TTL back by +1, hiding the fact that the packet was forwarded through your phone. So to the carrier, the traffic looks like it originated directly from the phone, not from a tethered device. That defeats one of the carrier’s tethering-detection methods."

There is a vague current popular idea that consulting LLM output as part of decision making is somewhat morally gray. This notion is not going to age well (or even ring true) in the near term. LLMs are simply a new way to store and use information and have their own warts, which can be said about all methods of information transmission.

LLM product managers: Show me what's in the context convenient to where I am prompting. Likely the user knowing and editing the precise context between requests will be a user task for a long time

I love this, it feels like something that should have always existed. great job

I don't think the author understands Amazon's AI product offering or what gen AI is doing internally at Amazon.

Strongly agree with this approach. Real world task specific LLM query workloads tend to be long lived workflows. They should be brokered and handled rather than wait on.

I fully discount the right to left thing.

There is not enough q/kdb full source code "out there" that would have made it into the LLM training data. It tends to be used in secretive environments and can be very site specific in convention. I bet a purpose built small fine tune with real + synthetic data would be enough to get something generating better Q code.

There is a tradeoff, they have to take immunosuppressive drugs for the rest of their lives.