// Computer Scientist · IBM · Islamabad, Pakistan
Syed Asad
Zaman
I'm a computer scientist — the specific problem changes more than the approach: reinforcement learning, differential privacy, and for the last two years, AI systems built to survive production traffic, not just a benchmark. Three of those are shipped and running real workloads at IBM.
Currently: applying that same instinct to enterprise asset data at IBM — read the system before touching it, build the smallest thing that actually fixes what's broken. Writing about what holds up once a system leaves the whiteboard.
// latest posts all posts →
- Your Degree Taught You to Build Software. Your Job May Ask You to Own the Problem. Forward Deployed Engineering is not just another job title. It points to a bigger shift in software engineering: from shipping code to taking responsibility for whether the problem was actually solved.
- Before We Fear AGI, We Should Ask If AGI Is Even Possible Before debating when AGI will arrive, there is a more basic question: do we actually know that artificial general intelligence is possible, or that today's LLMs are on the path toward it?
- The Big AI Lie: Software Engineers Were Never Going Away AI can write more code than ever, but Salesforce shows why writing code and replacing software engineers are two very different things.
- Single-Agent Is Not the Answer. It's the Current Shortcut. Why defaulting to single-agent architectures makes sense today, but treating it as a permanent principle is shortsighted.
- The AI Layoff Problem Is Not What You Think Why the AI layoff wave is a coordination failure that hurts firms too, and why UBI, capital taxes, and retraining can't fix it.