Sign In

Ariya Hidayat

LLMs: Amplifying human potential

Ariya Hidayat is a technology and business leader with extensive experience in the software industry.1 He is currently the Chief Architect at Hijra, a position he has held since July 2022.1 Ariya is also a Partner at Hyperjump since March 2018.1

His career history includes notable positions such as:

  • Chief Architect at Metabase (November 2020 - June 2022)
  • VP of Engineering at Shape Security (October 2013 - January 2018)
  • Engineering Director at Sencha, Inc (August 2010 - October 2013)
  • Staff Engineer/Manager at Qualcomm (December 2009 - August 2010)
  • Software Engineer at Trolltech (April 2008 - October 2009)1

Ariya is known for his contributions to open-source projects, including PhantomJS, Esprima, FastLZ, and KabarVirus.2 He is active in the tech community, sharing insights on JavaScript and software development.3 His LinkedIn username is indeed ariyahidayat.14

It's worth noting that while the query mentions Ariya as an Advisor at Ario, this position is not listed in the provided search results. The most recent information shows him as Chief Architect at Hijra.1

Highlights

Nov 17 · linkedin

During a recent #LLM meetup facilitated by Gather in Palo Alto, I gave a demo on #VectorSearch and LLM. It was an improved version of Pico Jarvis, now equipped to respond to questions based on content extracted from a PDF document (the popular "Chat with Your PDF" style). For this demo, the said document was a PDF export of the Wikipedia article detailing our solar system. The idea is to use an #embedding model to pinpoint relevant passages to the query, utilizing them as context for the LLM. Fortunately, #Mistral, the LLM I employed, flawlessly delivered results during the demonstration.

Note that the entire system is intended to be offline-first, there is no need to call any thirty party services for the embedding (#MiniLM with local transformers) and the LLM (I use #Mistral locally with #llama.cpp). Works like a charm on any decent computing device!

A last-minute addition made to the demo was the attribution of the answer. As you can see from the last part, Pico Jarvis happily told me that it discovered the distance to the closest star (note the support for multi-turn conversation) from a paragraph on Page 2. Checked it myself, and yup, it was spot-on.

For those interested, the complete code for the demo is accessible at https://lnkd.in/gP4V-Bee. Feel free to experiment with, modify, and provide feedback on the code. I look forward to our future LLM discussions!

Oct 3 · linkedin

Calling all software engineers who've dipped their toes into the world of ChatGPT and GitHub Copilot but are yet to explore the wonders of #LargeLanguageModels!

Are you curious about how LLM can supercharge your next project? Well, I've got some insights to share with you. Join me for an engaging event, hosted by Gather at their shiny office in Palo Alto, where I'll walk you through my journey of learning LLM and how to harness its potential to create something awesome.

No jargon, no buzzwords – just a straightforward, practical approach. We'll skip the convenience libraries and dive straight into the fundamentals. Together, we'll start from scratch with simple text generation and work our way up to crafting a weather-knowing bot!

Don't miss out on this casual session. Be there or be rectangular!

Related Questions

What are some of Ariya Hidayat's most notable projects?
How did Ariya Hidayat transition from software engineering to leadership roles?
What are Ariya Hidayat's key contributions to the tech industry?
Can you provide examples of Ariya Hidayat's work on JavaScript API design principles?
What is Ariya Hidayat's role at Ario, and how does it align with his previous experiences?
Ariya Hidayat
Ariya Hidayat, photo 1
Ariya Hidayat, photo 2
Add to my network

Location

San Francisco Bay Area