https://vlad-duda.com/ Recent content on Vlad Duda Hugo -- gohugo.io en-US Wed, 02 Oct 2019 15:59:13 -0400 https://vlad-duda.com/time/ Wed, 02 Oct 2019 15:59:13 -0400 https://vlad-duda.com/time/ Well for one, the first come first serve principle doesn’t always hold true within the landscape of software. Time is maleable within the computer science world. Misunderstanding time has led to space shuttle explosions, database exploits, incorrect gps direction, and plenty of other examples! For the past few years, I have devoted a significant amount of time to troubleshooting issues related with time. Time has a massive impact on our society based on how we: define, calculate, and understand time. https://vlad-duda.com/about/ Mon, 01 Jan 0001 00:00:00 +0000 https://vlad-duda.com/about/ About Enjoys hacking, writing, and andrenaline. An internet gentleman that builds digital things. A career consisting of startups, corporate tech, and research. Roles such as Software Engineer, DevOps, MLOps, Cofounder, and Consultant. When not overwhelmed; I enjoy sailing, boxing, reading, and spending time with friends & family. Subjects I’m interested in: Reverse Engineering Productivity Hacks Cutting edge software infrastracture Immigration/Legal tech policy eCommerce (from the technical & research side) Software wrk2-docker: Benchmarking Network Connections tor-exit: Monitor Tor Exit Nodes visc2ovpn: Convert file formats PandaCoin: AI/ML meets blockchain api-gateway-consul: Personal API Gateway magmi: Magento Mass Importer konsul: LLM for Immigration Advice Open Source OSSU: Open Source Society University tldr: Better than Google/SO Rancher: Centralized Kubernetes Control Plane (On-Prem) OpenRefine: Dealing with Dirty Data vuestorefront: Headless eCommerce PrivateBin: Zero-Trust PasteBin awesome-bash: bash widgets Fun PoC||GTFO: IYKYK Phrack Hubris Humility Index: Score high in both to win WTF Happened in 1971: Gold standard and Bitcoin The Machine Stops Dead Internet Theory: bots, bots, bots Monte Carlo: First document describing Monte Carlo Reverse Engineering LLMs Questions What Happens When? https://vlad-duda.com/blog/ai-cycle-insights-2016-2018/ Mon, 01 Jan 0001 00:00:00 +0000 https://vlad-duda.com/blog/ai-cycle-insights-2016-2018/ Drawing from the AI cycle of 2016-2018, there are some insights that can be developed: Importance of Strategic Planning: Many startups fell into the trap of assuming that AI development would be a linear process, and that increased computing power and data would naturally lead to improved AI capabilities. Infrastructure takes time to build (and time to sell to enterprises). Need for Sustainable Business Models: The failure of the “Human-in-the-loop” approach underscores the need for sustainable business models. https://vlad-duda.com/blog/genai-winnners/ Mon, 01 Jan 0001 00:00:00 +0000 https://vlad-duda.com/blog/genai-winnners/ As LLMs become more widespread, there are a few key groups of potential winners: Incumbents with existing distribution networks: Companies like Google, Microsoft, and Amazon can layer LLMs into their existing products and services to enhance the user experience. For example, Google Search could use LLMs to provide more comprehensive and informative answers to user queries. New companies that develop specialized LLM applications: There is an opportunity for new companies to develop specialized LLM applications that are tailored to the needs of specific industries or user groups. https://vlad-duda.com/blog/maker-time/ Mon, 01 Jan 0001 00:00:00 +0000 https://vlad-duda.com/blog/maker-time/ Time is the most important factor in creating high-performing software teams. Managers must carefully manage time, scope, and people to deliver successfully. Time: The total time the team spends on a problem. Scope: The functionality of the completed work. People: The number of people on the team and their skills. What teams do with their time is the primary factor in determining their output. Even if you hire the best engineers, they will not succeed if they spend all their time on the wrong things or are constantly interrupted. https://vlad-duda.com/blog/prompt-engineering/ Mon, 01 Jan 0001 00:00:00 +0000 https://vlad-duda.com/blog/prompt-engineering/ Prompting large language models (LLMs) is more of an art than a science, and it requires significant effort to develop useful prompts. However, prompt engineering is not a sustainable approach to querying LLMs, and it won’t be necessary in the future. Current prompt engineering methods rely on concatenation or templating, but these have limitations for handling complex prompts and require structured data inputs and outputs for best use. Prompt engineering today primarily relies on prompt concatenation or templating. https://vlad-duda.com/blog/future-of-llms/ Mon, 01 Jan 0001 00:00:00 +0000 https://vlad-duda.com/blog/future-of-llms/ The shift towards User-Friendly approaches of data Python has long been the go-to choice for many tasks involving data — from analyzing data to developing machine learning models. It’s been preferred largely because it’s easy to use across different computer systems, it doesn’t require rigid data type definitions, and it’s great for quickly trying out ideas, which is often needed in data-related fields. Anything that can’t be done in SQL is done in Python.