BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Palaner//palaner Single Event//EN
METHOD:PUBLISH
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:meetup-d62d37adea@palaner.app
DTSTAMP:20260802T201956Z
DTSTART:20260826T170000Z
DTEND:20260826T183000Z
SUMMARY:AWS AI In Practice 6
LOCATION:AutogenAI
DESCRIPTION:Saved in Palaner as: saved\n\nPalaner match: 29% · why: lots of
  people going\n\nWe’re delighted to welcome Anton Nazaruk\, CTO\, Cloud Co
 mbinator\, Ivaylo Iliev\, Partner Solutions Architect\, AWS and Silvia Leh
 nis\, Chief AI Officer\, UBDS Digital\n\nHere’s what they’re bringing:\n\n
 Seven ways to buy GPU compute on AWS. One wrong choice and the bill climbs
  while your training run stalls. Anton builds production-ready data and AI
  platforms at Cloud Combinator\, and he’s bringing the receipts - a pre-re
 corded walkthrough of a real distributed training run\, including a node f
 ailure\, replacement\, and full job recovery. Tonight he’s joined by Ivayl
 o\, giving us the complete blueprint: Capacity Blocks\, SageMaker HyperPod
 \, EFA networking\, FSx storage\, and Slurm or EKS orchestration\, with co
 st control designed in from day one.\n\nSilvia and her team at UBDS Digita
 l took an intelligent document processing solution on Amazon Bedrock and t
 ested everything - prompting\, model selection\, document handling\, and v
 alidation - to find out what actually moves extraction accuracy on messy\,
  real-world documents. Tonight she’s walking us through the results: which
  methods earn their keep\, which don’t\, and the trade-offs between value 
 and implementation effort.\n\nA big thank you to our sponsors Cloudscaler\
 , Rayo & The Scale Factory for making this event possible.\n\nProgramme:\n
 18:00: Arrival\, registration\n18:15: Talks start\n20:00: Networking with 
 food and a drink provided by the generosity of our sponsors.\n\nSession 1:
 \nFrom Zero to HyperPod: Choosing\, Operating\, and Cost-Controlling Distr
 ibuted Model Training Infrastructure on AWS with Anton Nazaruk & Ivaylo Il
 iev\n\nTraining large models on AWS is no longer just an ML problem - it i
 s a capacity\, infrastructure\, reliability\, and cost-control problem. Th
 is talk gives engineers a practical framework for choosing between EC2 On-
 Demand\, Spot\, Savings Plans\, Capacity Blocks\, SageMaker Training Jobs\
 , SageMaker Training Plans\, and SageMaker HyperPod.\nWe’ll look at when e
 ach option makes sense\, what tradeoffs they introduce\, and how to avoid 
 common mistakes around GPU availability\, quota planning\, interruptions\,
  and runaway cost.\nWe’ll then walk through a repeatable distributed train
 ing blueprint: compute fleet\, EFA networking\, FSx/S3 storage\, Slurm or 
 EKS orchestration\, observability\, checkpointing\, and failure recovery.\
 nThe session includes a demo-style walkthrough of launching a distributed 
 training job and showing how node failure and recovery should be handled i
 n a production-ready setup.\nThe goal is for attendees to leave with a cle
 ar mental model of how to run distributed model training on AWS reliably\,
  how to choose the right service or capacity model\, and how to make cost 
 and failure recovery part of the architecture from day one.\nLearning Take
 aways\n\nChoose between EC2 On-Demand\, Spot\, Savings Plans\, Capacity Bl
 ocks\, SageMaker Training Jobs\, Training Plans\, and HyperPod with a clea
 r framework for when each makes sense.\nBuild a repeatable distributed tra
 ining blueprint - compute fleet\, EFA networking\, FSx/S3 storage\, Slurm 
 or EKS orchestration\, observability\, and checkpointing.\nDesign cost con
 trol and node failure recovery into your training architecture from day on
 e.\n\nAnton Anton Nazaruk is CTO at Cloud Combinator\, where he works on c
 loud architecture\, AI infrastructure\, and distributed systems. He helps 
 teams design production-ready platforms for data and AI workloads on AWS\,
  with a focus on reliability\, cost control\, and repeatable infrastructur
 e patterns.\nIvaylo is a Partner Solutions Architect at AWS\, where he hel
 ps organisations optimise their cloud solutions and accelerate their digit
 al transformation journeys. With a focus on artificial intelligence and ma
 chine learning\, he works closely with AWS partners to develop innovative 
 solutions that address real-world business challenges. His expertise spans
  cloud architecture\, AI implementation\, and building scalable solutions 
 for customers across diverse industries.\n\nSession 2:\nBeyond Prompt Opti
 misation: Improving Accuracy in LLM-Based Document Extraction with Silvia 
 Lehnis\nVariable document formats make structured data extraction difficul
 t to solve with prompting alone. Especially when the documents include eve
 rything from images\, handwriting\, complex financial tables repeated mult
 iple times and without a particular form or naming conventions.\nThis talk
  explores how we optimised an intelligent document processing solution on 
 Amazon Bedrock by testing 17 different variants across prompting\, model s
 election\, document handling and validation. We will share the methods to 
 optimise an LLM based solution that apply to many use-cases beyond documen
 ts\, as well as the key trade-offs\, failure modes and design decisions th
 at had the greatest impact on extraction accuracy and reliability.\nLearni
 ng Takeaways\n\nHow to test and compare AI solution options\, including mo
 del selection\, experiment design and evaluation methods.\nHow the finding
 s shaped key design decisions and the final Amazon Bedrock solution releas
 ed into production.\nWhich methods can improve the performance of an LLM-b
 ased solution\, and the trade-offs between value and implementation effort
 .\n\nSilvia helps high-impact organisations use data and AI to reach their
  goals faster and safer. She has led global and national data and AI trans
 formations in sensitive environments across the public sector\, finance\, 
 energy and academia\, taking strategy through to implementation across peo
 ple\, process and technology - with solutions reaching up to 110\,000 user
 s and delivering £27m in savings over three years. She’s also a board memb
 er of the charity Care in Action.\n\nDo you have a story to share? If you 
 are interested in speaking at one of our events\, please check out our cal
 l for papers.\n\nWe are advocates for greater inclusion & diversity in UK 
 Tech and are especially keen to receive talk submissions from people in un
 derrepresented groups. If you are interested in speaking at a future meetu
 p but would like to discuss what to expect or need assistance\, please con
 tact our Inclusion & Diversity Lead Natalie Gray - graynataliej@gmail.com 
 or DM her @natjgray\n\nCheck out our website for more information about ou
 r community and our code of conduct. Remember to follow us @AWSUserGroupUK
  and on LinkedIn for the latest updates\, and you can find videos of our p
 ast meetups here.
URL:https://www.meetup.com/awsuguk-ai-in-practice/events/311421596/
CATEGORIES:palaner,saved
STATUS:TENTATIVE
END:VEVENT
END:VCALENDAR