Soon they were led to a hall that was huge and resembled a luxurious conference room. On the other side of the long rectangular granite table sat four people. One looked Indian, while the others were white foreigners.
"Sir, Mr. Aric More is here..." the girl who led them announced as soon as they walked in. The group of people, who seemed to be checking some documents, lifted their heads and saw Aric. All of them stood up, and the man leading them was easily recognizable.
Eric Schmidt, the current CEO of Google. In fact, there was another face Aric recognized. Eric stepped out from behind the chair and walked straight to Aric.
"You must be Aric... You have the same pronunciation as my name... Hahaha," Eric said. This made Aric smile too. Aric didn’t know much about this man who had led Google for a long time and was one of the reasons Google was so huge. Even now, it had started expanding at a rapid speed.
"It's nice to meet you, Mr. Schmidt," Aric said, shaking his hand.
"I gotta say... I knew you were young, but you look younger than I imagined," Eric remarked.
"I’m glad I was able to surprise you," Aric replied.
"You must be Aric’s father, Eklavya. I’m sorry if I mispronounced that..." Eric shook hands with Aric’s father, who was still trying to process what was going on. He subconsciously shook Eric’s hand. Eric could tell his father was confused and looked at Aric.
It didn’t take long for him to understand that Eklavya had no idea what was going on, so he introduced himself. Knowing the designation of the man he was speaking to, Aric’s father's heartbeat rose rapidly. He thought they would be meeting some heads of Google, but not the CEO.
"Hello..." Aric’s father replied, a little embarrassed.
He hadn’t expected the CEO of Google, and even Uncle Kurien was flabbergasted, secretly pinching himself to check if what was happening was real. Both of them felt like they were in autopilot mode.
"Please, take a seat," Eric said as he finally returned to his seat after Aric and Eric introduced everyone.
"Let’s get down to business since we’re done with the formalities. Shall we?"
"Of course," Aric replied.
"The image compression software you’ve introduced is at least 10 to 15 years ahead from our perspective, and our analysts estimate that in the next 50 years, there won’t be a better software, even as hardware advances," Eric said, his words filled with admiration for Aric's software.
If you encounter this story on Amazon, note that it's taken without permission from the author. Report it.
He had his analysts and ethical hackers look into the software, and they were both thrilled and scared by what they saw. There were no vulnerabilities whatsoever. They tried their best to understand the source code but couldn’t crack even a small portion of the inner workings.
Normally, a software program could never be completely unbreakable, and reverse-engineering the source code would usually provide at least some insights, giving them an edge in negotiations. But this was a complete black box.
This left Eric highly impressed, especially when he learned that the person behind the software wasn’t even 18 years old. At first, he refused to believe it, but when his team double-checked and confirmed that Aric was indeed the developer, Eric’s doubts faded once he met Aric in person. The confidence Aric exuded was enough to convince him that this was the real deal.
"Thank you for taking the time to meet with me today, Aric. Google is always on the lookout for cutting-edge technology, and I’ve heard some impressive things about your new image compression software. Could you tell me more about it?" Eric asked. This question was mostly for the benefit of the associates he brought. Although he had a background in computer science, what he had seen was high-level even for him.
"HyperComp leverages the current hardware available to achieve results as quickly as possible. Unlike traditional methods that rely on discrete cosine transforms (DCT) or wavelet transforms, HyperComp uses a combination of deep learning techniques and context-adaptive binary arithmetic coding."
Eric glanced at his colleagues, and they nodded in agreement. The first question had been to confirm that Aric was indeed the person behind the software and not just a poster boy. Knowing he was the right person, they were prepared to proceed further.
"That sounds intriguing. Can you elaborate on how it works and what makes it so advanced?"
"Certainly. Traditional methods compress images by breaking them down into smaller, more manageable pieces, then reducing redundancy within those pieces. HyperComp, however, uses a neural network-based encoder-decoder architecture that learns to represent image data in a highly efficient manner.
The process begins with the encoder, which analyzes the image and creates a compact, latent representation of the data. This representation is passed through a context-adaptive module that optimizes the compression based on the image’s content. Finally, the data is encoded using CABAC, which further reduces the size by efficiently handling probabilities and context modeling."
"That’s fascinating. How does HyperComp handle the trade-off between compression and image quality? Typically, higher compression ratios lead to a loss of detail and artifacts."
"Excellent question. HyperComp employs an adaptive neural network that constantly evaluates the visual importance of different image regions. This network, which we call the Adaptive Perceptual Network (APN), ensures that important details are preserved while less critical information is compressed more aggressively.
Moreover, HyperComp uses perceptual loss functions during training, which are designed to maintain high visual quality even at lower bitrates. This allows us to achieve a balance between compression ratio and image quality that is superior to traditional methods."
The answers Aric gave were too high-level even for some of Eric’s colleagues at this point. They could only vaguely understand and nod their heads.