About
Ensuring technology priorities are clearly connected to business goals, and that teams…
Activity
2K followers
-
Edwin Ng shared thisData Sharing Issues in a Microservice ArchitectureData Sharing Issues in a Microservice Architecture
-
Edwin Ng shared thisMicrosoft-sponsored Radius project aims to mitigate “limitations of Kubernetes” • DEVCLASSMicrosoft-sponsored Radius project aims to mitigate “limitations of Kubernetes” • DEVCLASS
-
Edwin Ng shared thisgood one... https://lnkd.in/gqMwwtav
-
Edwin Ng shared thisAkamai launches new cloud computing regions in Asia, Europe and the Americas | TechCrunchAkamai launches new cloud computing regions in Asia, Europe and the Americas | TechCrunch
-
Edwin Ng shared thisA rare inside-look at Cloudflare's massive networking backboneA rare inside-look at Cloudflare's massive networking backbone
-
Edwin Ng shared this17 free AWS cybersecurity courses you can take right now - Help Net Security17 free AWS cybersecurity courses you can take right now - Help Net Security
-
Edwin Ng shared thisHow to Prevent API Breaches: A Guide to Robust SecurityHow to Prevent API Breaches: A Guide to Robust Security
Edwin Ng commented on a post
2mo
very sorry to hear this Dato. May the LORD God Jesus Christ bring you healing and comfort during this difficult time. May His will be done in your life.
-
Edwin Ng liked thisEdwin Ng liked thisI reckon spec-driven development, agent governance and guardrails are starting to converge. As agents write more code and take more actions, reviewing everything manually gets harder. Requirements and policies need to become things software can actually check. Bend (https://lnkd.in/gKDSdGnf ) makes that tangible: developers define laws that code must satisfy, backed by checked proofs. Harmonic ’s Aristotle helps generate proofs. Both connect to the challenge described in Intent Formalization ( https://lnkd.in/g-GsVNyV ): translating what people want into precise, checkable specifications. Amazon Web Services (AWS) Bedrock’s Automated Reasoning checks bring a similar idea to generated answers, checking whether they follow defined policies. Research like AgentSpec ( https://lnkd.in/gUVRSVa4 ) extends specifications into runtime rules that can trigger interventions or block agent actions. NVIDIA ’s NeMo Guardrails provides configurable checks around conversations and tools. Different approaches, but increasingly connected: define the expected behaviour, establish boundaries, then check and enforce them. Scaling this remains hard. August’s Vero study tested agents generating code and proofs across multiple modules. The strongest tested agent solved 27 of 43 instances. Encouraging, with plenty still to figure out. And there’s a catch. Proving that a system follows a rule doesn’t mean we wrote the right rule. We can become impressively rigorous about getting the wrong thing right 🙂 Someone still needs to define what “right” means.
-
Edwin Ng liked thisEdwin Ng liked thisPress conference yesterday on Agentic AI ("Personalised Targeting" & "Assist") as new features that have been available on Jobstreet by SEEK platform, as well as press conference in June on Salary Pulse 2026, an exclusive research-based report provided to hirers for free. The latest press conference was followed with HR Networking with insightful Talk Show featuring 3 respected HR Directors: Bu Dina Sitopu, ACC, Pak Dr. Irwan Dewanto & Bu Agustina Samara. I would like to thank journalists from various media who attended press conference, HR Leaders from various companies who attended HR Networking, 3 Talk Show panelists who made fruitful discussion during the events and all Jobstreet Indonesia by SEEK team who have done great effort to make these events successful!
-
Edwin Ng liked thisEdwin Ng liked thisI have been building a RAG chatbot that answers questions from church sermon notes. What started as "let me see if I can make this work" turned into me making architecture decisions that would have been very easy to overcomplicate. I ended up making four choices that I found more interesting than the chatbot itself: • I built a custom Markdown parser because my sermon notes already have structure. Why throw structured data into a generic text splitter? • I used SQLite-vec instead of introducing a separate vector database because the corpus is still tiny. • I used a ContextVar for request tracing because the application is async and thread-local state would not give me reliable request correlation. • I used deque(maxlen=8) for chat memory instead of Redis or langchain's conversation memory before I actually needed it. There is also provider failover for both embeddings and text generation because a chatbot should not stop answering questions simply because a provider is down or rate-limited. None of these are "best practices" that I think everyone should copy. They are decisions that make sense for the system I have today. That is probably the bigger lesson I have taken from building this: I wrote about the decisions in more detail, including how the sermon notes are parsed, how the vector search works, what happens when the application gets multiple workers, and the tradeoffs behind the provider fallbacks. I have linked the full article below. https://lnkd.in/d_wcAHA4Engineering Decisions I Almost Overcomplicated When Building A ChatbotEngineering Decisions I Almost Overcomplicated When Building A Chatbot
-
Edwin Ng liked thisEdwin Ng liked this25 years is a milestone. But more importantly, it is a reminder of the responsibility we carry. Over the past week, we came together as Allianz Malaysia to mark our 25th anniversary through our townhall, company-wide team building and the premiere of a documentary we made that tells the story of our 25-year journey in Malaysia. For me, this anniversary is deeply personal. I joined Allianz Malaysia in 2004 and over the past 22 years, I have had the privilege of serving in different parts of the business. Looking back at old photographs, milestones and stories throughout this past week reminded me of why I chose Allianz Malaysia then, and why I continue to choose it today. It is because this company has never been built by numbers alone. It is built by our people. Yes, we are proud to be a leading insurer in this country. But what makes me most proud is knowing that behind every policy, every claim, every family supported, every business protected and every community we give back to, there are people in Allianz Malaysia who care deeply about doing the right thing. And as we celebrate 25 years, I also believe we are only beginning our next chapter. We are not here to rest on what we’ve achieved but to continue growing, doing better and protecting more Malaysians. Here’s to everyone who has been part of the Allianz Malaysia story. Whether you have been here from the beginning or joined us recently, you are on this incredible journey with us. Together, let us continue to secure the future of the people, businesses and communities who place their trust in us.
-
Edwin Ng liked thisEdwin Ng liked thisMy dad Dr. Khalid Abulmajd used to tell me this when I was young: “Listen to understand, not to reply.” At the time, I didn’t fully understand what he meant. It was only as I grew older that I began to realize how powerful those words really are. It sounds simple, but it can truly change the way you see people and lead them. I often notice how difficult it is for us to listen without already preparing our response. We bring our own biases, emotions, and opinions into the conversation. Sometimes, we even judge the message based on the person’s personality, background, or beliefs. For me, the best way to practice this is to imagine I’m reading a book. I don’t know the author, their personality, or their background. I only have the words in front of me. I let those words reach my mind without filters, take a moment to understand them, and then respond in a way that adds value. It’s not always easy, but I believe this is one of the most valuable habits a leader can develop. Because sometimes, the best thing you can do for someone is simply to understand them before trying to be understood.
-
Edwin Ng liked thisEdwin Ng liked this“Hi Sean, I’m not actively looking for a new role, but I’m always open to a chat.” This might seem like a standard reply. Polite, noncommittal and deliberately vague. But after 14 years in IT recruitment, I’ve found very few responses carry as much weight as this one. I’ve learned not to assume it means someone wants a new job or that they’re necessarily ready to jump ship. A reply like this often has multiple meanings. - What am I worth right now & where am I positioned in the market? - Who are you and what do you know about my field/industry? - Can I trust you enough to tell you what’s really going on? - Are you credible enough to understand my position? - Projects have stalled and I can’t see a clear runway. - There’s a lack of vision and I’ve lost faith in Leadership And yes, sometimes they absolutely want out, they’re just not ready to say it yet. I’ve come to realise a response like this can mean so much, but there needs to be a level of trust before someone will really open up. You normally don’t find out which one it is until you’ve earned the conversation.
-
Edwin Ng liked thisEdwin Ng liked thisHappy 69th National Day, Malaysia! 🇲🇾 Being on a career break these past months has not only given me the time to recharge, but to also focus on the important things in life - being my wife’s caregiver as she recovers from her ankle surgery, chatting with my son about life and airplanes while sending him to school, and being able to invest time on my continued learning in this age of AI disruption. Being back in Malaysia after 2 years away, I remain hopeful that as my beloved Malaysia progresses, her rakyat will also continue to grow together as one people - to see past the divisions and to understand that our diversity is our nation’s greatest strength, not weakness. Thank you to my friends & acquaintances who have checked in on me with words of encouragement. I am reminded of Romans 5:3-5, which teaches that suffering produces endurance, which builds character and leads to a secure hope rooted in God's love. To my fellow Malaysians, Selamat Hari Kebangsaan and to everyone else, have a great week ahead!
-
Edwin Ng liked thisEdwin Ng liked thisI’ve led engineering teams across 7 countries. The best teams weren't built by process. They're built by trust. I've built teams from scratch to 100+ engineers. I've led 250+ global engineers. I've mentored and grown engineers across the United States, India, China, Europe, and the U.K. One of the biggest lessons I learned is this: A high-performing team does not start with process. It starts with the environment. Years ago, I inherited an underperforming team on a critical customer program. The customer was frustrated. Trust was low. The team was under pressure. And we were close to losing the account. There was no time for a long transformation plan. We had weeks. So I focused on the fundamentals. Create safety. Rebuild trust. Clarify roles. Listen deeply. Remove confusion. Respect the people doing the work. And make sure every engineer understood why their contribution mattered. The shift was not instant. But it was visible. People started speaking up. Problems surfaced earlier. Leaders stopped managing from fear. Engineers started solving across boundaries. The customer saw the change. And two months later, we launched. That experience changed how I think about team performance. Because high-performing teams aren't just more skilled. They're safer. Clearer. More accountable. More connected. More willing to tell the truth before the problem gets bigger. That's why the frameworks here matter. They're not academic ideas. They are leadership tools. They help leaders build the kind of environment where people can do their best work. If you want to build a high-performing team, start here: 1. Create psychological safety. People need to feel safe enough to speak up, challenge ideas, and surface risks early. 2. Build real trust. Trust is not built in a meeting. It is built when leaders listen, follow through, and respect the team. 3. Clarify the work. Confusion slows execution. Clear roles, goals, and decision paths create speed. 4. Develop shared mental models. Teams perform better when they understand the same goal, the same priorities, and the same operating rhythm. 5. Give better feedback. Feedback should improve performance, not create fear. 6. Align around the outcome. When people know what matters most, they make better decisions without waiting for permission. 7. Lead with consistency. The team watches what you reward, what you tolerate, and what you model. The best teams I've led weren't perfect. But they were honest. They were committed. They trusted each other enough to solve hard problems together. And that's where performance changes. If you are ready for your next leadership chapter, let’s build the strategy to get you there. Executive advisory conversation: 🔗 https://lnkd.in/ePaf--N9 🔁 Share this with a leader with a team they want to inspire. ➕ Follow Stephanie Hills, Ph.D. for more on leadership, executive growth, and AI insights.
-
Edwin Ng liked thisEdwin Ng liked thisCongratulations to UOB Malaysia Chief Executive Officer, Datuk Ng Wei Wei, on being conferred the Darjah Mulia Seri Melaka (D.M.S.M.) which carries the title Datuk. This recognition reflects her leadership in advancing sustainable growth, supporting businesses and communities, and strengthening regional connectivity. It also underscores UOB Malaysia’s longstanding partnership in contributing to the development and prosperity of the country and the state.
Experience & Education
Other similar profiles
Explore more posts
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More