Suggestions
Achin Kumar
Building Technology Products of the Future | IoT | AI | Beauty-tech | Health-tech | Sports-tech | Behaviour Psychology
Achin Kumar is the Co-founder and CTO of Skin Dossier, a company focused on technology products in various sectors including beauty-tech. He has a strong background in product design and development, having worked with notable brands such as L’Oréal and Warner Brothers.
Professional Background
-
Current Role: Co-Founder & CTO at Skin Dossier, where he is involved in creating innovative solutions in the beauty technology space.
-
Previous Experience: Achin has held several significant positions:
- Co-Founder and CXO at Roro Solutions (since May 2016), a digital agency specializing in branding and development of mobile and web products.
- Head of Tech at Yura App (since March 2020), focusing on improving agency-client relationships through a streamlined platform.
- Product Designer and Developer at Vinsol (2011-2016), where he conceptualized and developed successful products for various clients.
Education
Achin Kumar holds both a Bachelor's and a Master's degree in Computer Science from the Symbiosis Institute of Computer Studies and Research, completed between 2006 and 2011.
Skills and Expertise
He is proficient in UX/UI design, mobile and web development, and is recognized for his ability to build products that address real-world problems. His experience spans across multiple technologies and platforms, making him a versatile figure in the tech industry.1
Highlights
An astounding high average of ~125 new research papers on LLMs are getting published EVERYDAY, on arXiv.com alone! 🤯
Too much is being done by too many people, too quickly.
It’d be insane to even try to keep up with all of it. It’s a giant wave of disruption. 🌊
But is it all relevant for someone building LLM apps? Not much, at least not immediately.
The key is to not get distracted.
For instance, don’t get blinded by GraphRAG, without trying out basic or corrective RAG.
If you’re building in Gen AI space, you’re in a hot soup. You might have a chance to fundamentally disrupt one industry or the other.
Focus on solving a real problem, rather than getting constantly swayed by the giant wave.
———————————————
So, should you completely ignore the papers? NO. You should definitely keep an eye, at least on the surface level papers. Stay excited and informed.
I’ll be regularly sharing my thoughts on cherry-picked, interesting papers. I'll also point out their impact on LLM app development. Follow me for that.
BTW, the high volume of research papers is a good sign. It’s the reasoning behind "AI will create more jobs than it would take away" notion. Will be writing about this in detail soon.
Here is a song to rant about this information-overloaded Sunday 🎵
https://lnkd.in/ghFj2VTA
#ai #llms #GenerativeAI #datascience #rag
TRUST IS THE KEY, whether it’s a human brain or a Deep Learning model. A huge blocker in acceptance of Generative AI applications in domains such as healthcare, finance and legal is its lack of explain-ability.
I talked to users and professionals in these domains and found out that they perceive GenAI as a black-box, and that’s a blocker.
LLMs are deep learning models and have such a complex neural network, that some layers of it are so complex that they can’t be interpreted. They’re called hidden layers.
It’s almost impossible to trace back the logic beyond the hidden layers. This is where logic turns into psychology. And that’s literally a black box. 🗃️
But then, exactly so is the human brain, only much more complex, with many more hidden layers.
When your doctor says you’ve acute bronchitis and punches you with a pack of pills, that’s also a response coming out of a black box. We take it or reject it based on two factors.
☝️How much do I trust the doctor, based on my past experience and general trust.
✌️How clearly can the doctor explain why he thinks what he thinks. (How much do I trust the explanation?)
That’s how a future LLM based professional Agentic-system will be weighed in trust.
———————————————
LLM interpretability / explainability is a critical field of study that some researchers are actively working on.
Check out this research paper on Explainable | Interpretable AI. https://lnkd.in/gTRJf3_A
There are ways in which GenAI applications are moving past this blocker. I’ll talk about them in another post soon.
#ai #llms #GenerativeAI #datascience #deeplearning #machinelearning
