ASI2024 02 6 Sourcing strategy скачать в хорошем качестве

ASI2024 02 6 Sourcing strategy

   • ASI2024 02 6 Sourcing strategy      • 2 6  Sourcing strategy   The final lecture in series on AI architecture. It shifts from technical implementation to a business-oriented perspective on sourcing strategies for IT and machine learning projects. The lecture addresses three critical questions for any modern entrepreneur or AI architect: 1. Custom vs. Out-of-the-Box Solutions The recommended strategy is to prioritize off-the-shelf solutions to minimize risk and accelerate time-to-market. Prioritization Strategy Out-of-the-Box: Use a ready-made product if one exists. Adopt/Modify: If a perfect fit isn't available, find a solution that can be adapted. Custom Build: Build from scratch only if no other options exist. When to Choose Out-of-the-Box Cost and Speed: The total purchase cost is lower, and it allows for much faster market entry (e.g., weeks vs. years). Universal Processes: For common business needs like HR management where processes are standardized. Proven Solutions: If market leaders already exist for the specific problem. When to Build Custom Competitive Advantage: The area is core to your unique business value. High Adaptation Costs: If modifying an existing product is more expensive than building it. Deep Business Knowledge: When external providers lack the sophisticated understanding required for your industry (e.g., niche internet marketing). Time and Delivery: When suppliers cannot guarantee delivery within your required timeframe. 2. Open Source vs. Commercial Solutions This choice often involves weighing community-driven freedom against reliable industry support. When to Choose Open Source Community Health: The code is well-supported by an active community (e.g., high stars and frequent commits on GitHub). Large Player Support: It is backed by a major company (e.g., MLflow supported by Databricks). Total Cost of Ownership (TCO): When the overall cost, including maintenance, is lower. Source Code Ownership: Essential for startups that need to prove ownership of their IP for future acquisition. When to Choose Commercial Industry Standards: The product is a standard in your field (e.g., specific spreadsheet software). Reliable Support: You need guaranteed long-term support that won't disappear. Developer Costs: Open source isn't "free"—sometimes the cost of hiring specialized developers to maintain it exceeds the price of a commercial license. Legal Compliance: To ensure you aren't infringing on copyrights or security protocols. 3. On-Premise vs. Cloud Hosting This decision is increasingly influenced by data privacy and low-latency requirements. When to Maintain On-Premise Full Control and Privacy: Essential for sensitive data (e.g., healthcare/GDPR) or preventing document leakage to external providers. Data Transfer Costs: To avoid high "ingress" and "egress" fees when frequently moving large volumes of data. Low Latency: Critical for real-time manufacturing and IoT plants where cloud response times are too slow. Operational Stability: When a third-party provider cannot guarantee the required level of uptime. When to Choose Cloud Business Focus: Allows you to focus on your core business rather than infrastructure management. Scalability and Flexibility: Ideal for growth and seasonality, particularly with "serverless" options like Google Cloud Run or AWS Lambda where you pay only for what you use. Reliability: Provides continuous, professional support for the infrastructure.