1. Analyze problems and design GenAI solutions
-
Work with business units to receive requirements, assess GenAI application capabilities and propose suitable technical approaches.
-
Analyze use context, input/output data and risks to design feasible, secure GenAI architecture.
-
Select models, platforms and deployment methods (on-premise, private cloud, hybrid, API service) appropriate for each problem.
2. Develop and deploy GenAI services
-
Design and deploy GenAI solutions on open-source model platforms or external API services (e.g. OpenAI, Gemini or equivalent), complying with the Bank's security policies and risk management.
-
Package GenAI applications as services (API/microservice) for integration with business systems, internal portals and digital platforms.
3. Develop RAG (Retrieval-Augmented Generation) solutions
-
Design and deploy RAG pipelines, including collection, preprocessing, indexing and knowledge retrieval from internal data.
-
Combine LLM with semantic search/vector database to ensure context-appropriate responses, alignment with internal data and hallucination reduction.
-
Coordinate with Data Governance and BI to ensure data used for RAG is classified, authorized and controlled according to regulations.
4. Develop AI Agents and intelligent automation
-
Design and deploy AI Agent/Agentic Workflow to execute task sequences from request reception, retrieval, analysis, response to action execution.
-
Integrate GenAI with workflow automation tools (e.g. n8n or equivalent platforms) to automate internal processes and support repetitive tasks.
-
Standardize Agents and Workflows for reuse, scaling and deployment across the Bank.
5. Monitor, control and improve GenAI applications
-
Establish monitoring, logging and quality assessment of GenAI applications on accuracy, context appropriateness, hallucination, performance and latency.
-
Track model usage costs, system resources and optimize performance and operational costs.
-
Improve prompts, pipelines and Agent logic when data or business requirements change.
-
Ensure GenAI applications meet requirements for security, authorization and input/output data control.
6. Standardize architecture and share GenAI knowledge
-
Participate in building, standardizing and developing shared GenAI/Agentic AI architecture for the Bank.
-
Document solutions, pipelines, configurations and best practices.
-
Share knowledge, provide technical guidance and support implementation and scaling of solutions across the Bank.
