Couverture de EP12 : Forge & Perform AI Model Customization at Scale | AWS for AI Podcast | Amazon Web Services

EP12 : Forge & Perform AI Model Customization at Scale | AWS for AI Podcast | Amazon Web Services

EP12 : Forge & Perform AI Model Customization at Scale | AWS for AI Podcast | Amazon Web Services

Écouter gratuitement

Voir les détails

Offres de saison | 0,99 € par mois pendant les 3 premiers mois

5,99 €/mois par la suite. Des conditions s’appliquent. Annulation mensuelle possible.
À propos de ce contenu audio
Join host Hamza Mimi for a deep conversation on model customization with two guests tackling it from opposite vantage points. Oliver Gracer, Head of Data & Insights at Perform AI (formerly Parcel Perform) — the Singapore-based delivery experience platform processing billions of shipments across 160+ countries — shares how his team fine-tuned Amazon Nova and achieved something counter-intuitive: a smaller model that outperformed larger ones on accuracy, cost, and speed. Joining him is Karan Pandharkar, Principal Product Manager for Forge and model customization at AWS, who reveals the patterns, anti-patterns, and tools he sees across customers worldwide.Together they unpack why customization is an economics question before it's a technology question, and give a practical roadmap for teams moving from prompting to production customization. This episode covers critical topics including:• Why and when to customize vs. prompt engineer, RAG, or build an agentic harness• Building a robust evaluation framework and defining what "good" looks like• The ROI conversation: fine-tuning cost vs. inference cost (and talking to your CFO)• The hallucination problem and how fine-tuning tackled it• Pre-training, mid-training, fine-tuning, and reinforcement learning explained• Data quality over quantity — and how little data you may actually need• Catastrophic forgetting and the role of general-purpose data mixing• LoRA / parameter-efficient fine-tuning and the pay-as-you-go economics of serving• Customizing with SageMaker AI, Amazon Bedrock, and Nova ForgeWhether you're a startup, enterprise, or public sector organization running high-volume AI workloads, this episode offers invaluable guidance on building scalable, cost-efficient, production-ready AI solutions in today's rapidly evolving landscape.OUTLINE:00:00:00 : Introduction00:00:43 : Meet Oliver — Perform AI00:03:34 : Meet Karan — AWS Forge00:07:04 : Why bother customizing when frontier models keep winning?00:11:51 : The only 3 reasons to customize: quality, speed, cost00:13:58 : Does the math actually work? The ROI reality00:19:47 : When you should NOT customize00:22:11 : How a small model beat the giants in production00:25:36 : Making the model stop hallucinating00:28:04 : Pre-training to RL: the 4 stages demystified00:31:39 : You need way less data than you think00:39:29 : Catastrophic forgetting (and how to avoid it)00:44:08 : Why evals matter more than your model00:48:48 : Prompt vs. fine-tune vs. agent — it's not either/or00:50:34 : How to actually do it on AWS00:56:00 : From notebook to production00:56:15 : Quickfire round01:01:16 : A CTO's playbook for customization01:04:53 : Closing remarksTo learn more about Perform AI : https://bit.ly/4ywZ6OMParcel Perform case study blog : https://aws.amazon.com/blogs/machine-...
adbl_web_anon_alc_button_suppression_t1
Aucun commentaire pour le moment