AI Agents Foundation for Engineering Teams: Memory, Skills and MCP
- 2 days (16 hours, including breaks)
- Advanced
In short
What is the AI Agents Foundation for Engineering Teams: Memory, Skills and MCP program?
Build an agent prototype around a recurring engineering task. Work through the reasoning-action loop, context and memory design, reusable skills, Model Context Protocol connections, orchestration, and technical governance.
What your team will be able to do
- Implement a basic observe-plan-act loop with structured tool calls and stopping conditions.
- Design retrieval and persistent memory for a defined engineering task.
- Build a reusable skill package and connect a sandbox agent to an MCP server.
- Choose when a single agent is enough and when coordination is justified.
- Present a working prototype and peer-reviewed technical design.
No obligation. We reply within one working day, and your details stay confidential. Privacy
Who runs it
This program is run by industry practitioners who have shipped AI inside regulated enterprises across banking, government, manufacturing, and telco. It is led by Peter Kua, our AI and data science lead, a core contributor to Malaysia’s National Big Data Analytics Framework under MDEC and co-author of Malaysia’s AI Adoption Blueprints in 2019 and 2020. You are learning from people who have built this in production, not read about it. Peter’s commentary on AI adoption and workforce capability has appeared in The Star, theSun, Singapore Business Review, e27, and DagangNews, and on RTM Berita’s Tech Today.
Course outline
Participants build one prototype around a real recurring task from their team. Each module adds a component, from the minimum viable agent loop to a technical walkthrough.
Day 1: Architecture, context, and memory
Design the observe-plan-act loop and structured instructions. Build a retrieval approach that fits a context budget, then separate session state from persistent memory. The day ends with a peer review of the architecture and memory design.
Day 2: Skills, connections, and governance
Create a reusable skill, connect the prototype to a sandbox MCP server, and decide whether the task needs one agent or coordinated roles. Apply least-privilege access, sandboxing, logging, and cost limits before assembling the final prototype.
Entry requirements
Participants should be comfortable with Python or JavaScript and TypeScript, the command line, REST APIs, JSON, and Git. Bring a laptop and one recurring engineering task suitable for a sandbox prototype.
Common questions
Is this program HRDC claimable?
Yes. It is claimable through HRD Corp under SBL and SBL-Khas. GrowthPro Asia is an HRDC-registered provider, and we handle the claim paperwork so it stays audit-ready. For most registered employers the levy covers most or all of the cost.
Is this hands-on or theory?
Hands-on. These are build-along programs: your team works in real tools on real datasets and leaves with working code and a project it can reuse, not a slide deck. We introduce concepts only to support what you build.
What do we need to know before we start?
It depends on the level. Beginner programs start from fundamentals and assume no prior coding. Intermediate and advanced ones assume working Python or data experience. Each program states its level, and we help you place your team correctly.
Is it delivered in-house or as a scheduled cohort?
Both. Most teams run it in-house over two days on their own stack and data, which is where the skills stick. A scheduled cohort is available when you want a smaller group to learn alongside peers from other organisations.
How is the cost set?
We do not publish a fixed price. Cost depends on in-house versus scheduled delivery, group size, and how far we customise to your stack and data. It is HRD Corp claimable, so for most registered employers the levy covers most or all of it. Tell us your group size and format for an indicative quote.
Who should take this AI agents course?
It is for technical staff who will design, build, or operate agentic systems, including software engineers, platform engineers, technical leads, and solutions architects.
What technical experience is expected?
Participants should be comfortable with Python or JavaScript and TypeScript, command-line work, REST APIs, JSON, and Git-based workflows. Prior use of an AI coding assistant is helpful but not required.
What is the capstone?
Each participant brings a recurring engineering task and builds a prototype through the course, then presents it for peer review with a technical design and a key risk.
Related
- AI Capstone, the pilot a roadmap leads into.
- HRDC-claimable AI training, how the claim works.
- AI consulting, strategy, governance and pilots.
- Devin AI for Autonomous Software Engineering and Agentic Orchestration, applying task briefs and review controls to autonomous coding work
- Loop Engineering for Agentic AI, connecting agents into scheduled, governed workflows
- Advanced GenAI Systems: RAG, Fine-Tuning and Internal Copilots, a related program.
- Agentic AI for Workflow Autonomy, a related program.
Turn AI strategy into a working pilot, and prove the ROI.
Start with the free 3-hour AI Strategic Briefing. See agentic AI run real workflows, and leave with an automation roadmap, a prompt library, and a pilot playbook.