AI SW Team Manager

Key Responsibilities 

1. Enterprise GenAI & Agent Foundation (The Core) 

• Unified Agent Gateway: Architect a next-gen gateway that goes beyond simple LLM routing. Design the Agent Router capable of decomposing complex user intents and dispatching them to specialized agents (Coding Agents, HR Agents, Supply Chain Agents) or foundational models based on cost/performance/capability. 

• Agentic Runtime & Orchestration: Define the standard protocols for agent-to-agent communication, tool registration, and state management. Build the 〝operating system〝 where agents live, execute, and collaborate. 

• Generative UI (GenUI): Lead the architecture for Dynamic Interface Generation. Move beyond static chat boxes; design systems where the AI can generate bespoke UI components (charts, forms, interactive widgets) on the fly to best present data to the user (e.g., using Vercel AI SDK concepts or server-driven UI). 

2. Agentic RAG & Multi-Modal Intelligence 

• Agentic RAG Architecture: Shift from passive retrieval to Agentic RAG. Design workflows where agents actively plan queries, critique their own search results, and iteratively reason over data before answering. 

• Multi-Modal Data Strategy: Architect pipelines that ingest and retrieve across modalities—text, technical schematics (images), and code repositories. 

• Dual-Database Strategy: Implement a hybrid memory architecture combining Vector Databases (for semantic similarity) and Graph Databases (for structural relationships/GraphRAG) to ensure deep, context-aware retrieval. 

3. Generative BI & Data Products 

• Semantic Layer & Text-to-Insight: Bridge the gap between unstructured questions and structured Enterprise Data Warehouses. Architect the semantic translation layer that allows agents to reliably query SQL/NoSQL databases without hallucinating schema. 

4. Solution & Enterprise Architecture Strategy 

• The 〝Big Picture〝 Integration: Act as the bridge between GenAI innovation and legacy Enterprise IT. Ensure the platform integrates seamlessly with existing Identity Providers (SSO/IAM), ERPs (SAP), and PLM systems. 

• Blueprinting & Standards: Own the technical roadmap and architectural decision records (ADRs). Define the 〝Golden Paths〝 for internal development teams to build their own agents on top of our platform. 

• TCO & Governance: Architect for cost-efficiency (token optimization, caching strategies) and compliance. Define the patterns for data privacy and 〝Human-in-the-Loop〝 validation within agent flows.

1) 設計與建置 IT 客服 AI 支援方案 (如：服務品質監控、智慧助理)。 

2) 應用 IaC (基礎設施即程式碼) 原則，管理終端設備及用戶問題排除。 

3) 分析全球 IT 客服營運數據，優化服務流程及使用者體驗。 

4) 持續研究與導入新技術趨勢並提出改善建議 (如：無密碼驗證、WoA、AI PC)。

At MediaTek, we are at the forefront of enabling device+cloud hybrid architectures for generative AI. On-device AI offers significant advantages, including low latency, enhanced security, privacy, and personalization through fine-tuning to meet specific user needs. A critical component of this technology is the AI compiler, which involves code generation and optimization to maximize runtime performance and minimize memory footprint. We are seeking talented individuals to join our team and contribute to this cutting-edge field. 

Key Responsibilities: 

1. Develop and optimize AI compilers for on-device applications. 

2. Conduct performance analysis and implement optimizations for runtime efficiency and memory usage. 

3. Work closely with customers to understand their needs and provide tailored AI solutions.

【職務概述】 

我們正在尋找具前瞻視野與創新精神的工程領導者，加入生成式 AI 應用平台團隊，打造企業級 GenAI 平台，協助內部團隊快速開發 AI 應用與智能代理，推動數位轉型。您將領導團隊建構關鍵能力，包括 RAG 框架、多代理分析平台與 HyperAutomation 模組，並推進多代理協作、模型路由、LLMOps 標準化與負責任 AI 治理。 

【主要職責】 

策略與領導：帶領跨領域團隊，制定技術藍圖與 KPI，導入 RAG 2.0、知識圖譜、AI Agent 協作等技術。 

平台研發：設計與維運 RAG 框架、多模態分析 Agent、AI 自動化模組，導入 MCP、LangGraph 等前沿技術，強化平台效能與治理。 

利害關係人協作：作為技術策略夥伴，協助事業單位導入 GenAI，與資料、業務、安全團隊共創 AI 解決方案，推動開發者生態與負責任 AI 架構。

職缺說明 

我們是聯發科技IT MLOps平台團隊，致力於打造高效、可擴展的MLOps平台，提升企業AI模型開發、部署與運維自動化能力。團隊積極導入最新技術並持續學習成長，尋找具備MLOps平台架構設計與實踐經驗的菁英，一同推動公司AI數據驅動轉型。 

歡迎對雲端/地端AI平台、機器學習自動化與大型系統架構有熱情的你加入我們！ 

工作內容： 

1 規劃並設計企業級MLOps平台架構，負責專案規劃、目標訂定、技術選型與落地執行。 

2 與跨部門團隊AI合作，協助把商業問題拆解為可自動化實現的AI工作流，支撐AI模型端到端生命周期管理。 

3 主導平台核心模塊建置：模型訓練/部署管線、CI/CD for ML、資料流通與版本控管、模型監控與資產管理等。 

4 推動企業級MLOps標準化、流程自動化與最佳實踐，優化模型運營效能、穩定性及安全性。 

5 參與平台持續優化與技術創新，導入新技術·新工具如Kubeflow、MLflow、Vertex AI等，提升平台功能。 

6 協助建立團隊技術文化，分享MLOps相關知識、推動內部技術培訓與交流。

• Define GPU compiler software architecture and interfaces.  

• Development/implement GPU compiler pipeline, linking and various optimizations/transformations.  

• Collaborate with Driver team, HW team to implement new API & HW features.  

• Collaborate with Driver team, HW team to improve/tune performance & power consumption.  

• Execute & deliver to meet milestones/schedules.  

• Analyze and debug code generation issues.  

• Construct reliable & trustable relationships across teams internally & externally.