Generative AI - What Business needs to understand
Building LLM from scratch
Features of open source LLMs (Mistral 7B, Meta Llama 2, Google Gemma, Microsoft Phi-2) & Proprietary LLMs (GPT4-Turbo, Gemini 1.5, Claude 3)
Subword Tokenizers and Embedding models with demos
Vector database design patterns (SingleStoreDB, Others)
andPrompt Engineering with 1 Open source LLM, 1 proprietary LLM (Demo)
(Reinforcement learning & LLM Alignment - RLHF, RHAF, DPO)RHAF, and
Designing / Architecting Finetuning strategies
Finetuning with parameter efficient methods- Concept with Demo
Review & doopen-ubt clearing session on Lab: Finetuning with parameter efficient methods (Open source Mistral/Llama series or a smaller LLM)
GCP, andDesigning/ Architecting production grade Retrieval Augmented Generation (RAG) pipelines on AWS, Azure, GCP, Databricks
Advanced & Modular RAG with 2 case studies
Review & doubt clearing session on Lab: RAG (to be specified by Saravana, if needed) - working session
FinOps for Generative AI (Example of AWS Cloud tooling)- Concepts, Architectural perspectives & how to use
Architectural learnings from domain adapted llms- BloombergGPT, FinGPT, DocLLM JPMorgan, Contractassist - Thomson Reuter labs, etc
Working with Huggingface transformers & artifacts
Designing/ Architecting LLM based apps with Langchain & Llamaindex
Designing/ Architecting production grade data pipeline- Ingestion, pre-processing, wrangling (Azure-Databricks or GCP)
Review & doubt clearing session on Lab: Building data pipeline- Ingestion, pre-processing, wrangling (Azure-Databricks or GCP)
Designing/ Architecting with Mixture of Experts (MOE) models, multi-modal LLMs (Gemini Ultra, 1 open source) & Small language models
Performance benchmarks and evaluation of LLMs: more time needed with architect perspectives
Deploying & Scaling LLMs on Azure-Databricks. (Specialized VMs/ML Accelerators, LLM Orchestration, Model serving included)
Review & doubt clearing session on Lab: Deploying a smaller LLM (similar to Phi-2/GPTJ, GPT-Neo) on Azure-Databricks/ AWS
LLM Security & Governance- What an AI Architect needs to know for designing secure solutions
Post production observability and model optimization (incl handling model drift, DLP)
AI Gateways like Javelin (How to use them to build enterprise-grade FM-based apps)
AI Regulations, Governance & Responsible AI (with upto 4 customer cases)- EU AI Act, UK, US
Lab: Finetuning with parameter-efficient methods (open-source Mistral/Llama series or a smaller LLM)
Lab: Retrieval Augmented Generation (Azure/ Azure-Databricks/ AWS)
Lab: Buildpre-processing, anding data pipeline- Ingestion, pre-processing, wrangling (Azure-Databricks or AWS)
Lab: Deploying a smaller LLM (similar to Phi-2/GPTJ, GPT-Neo) on Azure-Databricks/ AWS
Digital Global FZE, SAIF ZONE, UAE
UAE : +97144192650 | USA : +19294592657
Copyright © 2024 Digital Gurukul - All Rights Reserved.
Powered by Digital Global
We use cookies to analyze website traffic and optimize your website experience. By accepting our use of cookies, your data will be aggregated with all other user data.