Amburi Roy
Think beyond the pattern.
I’m a software engineer and researcher exploring the assumptions, trade-offs, and failure modes behind modern software systems—from distributed systems and architecture to engineering effectiveness and AI-driven development.
Software engineering is full of patterns, best practices, and increasingly AI-generated answers. My writing and research examine the assumptions behind them—where they work, where they fail, and what happens when real systems meet the theory.
Experience across
Feature Research
2026 IEEE International Conference on Human-Machine Systems (ICHMS)
When Human-in-the-Loop Fails: Accountability Across AI Adoption Levels in Human–Machine Systems
DOI: 1109/ICHMS69701.2026.11602198
This paper examines the concept of Human-in-the-Loop (HITL), which is widely used to promote safety, responsibility, and trust in intelligent systems but lacks a clear operational definition. It argues that human involvement alone does not guarantee accountability, as HITL implementations differ substantially in decision authority, timing, information access, and causal influence.
Talks
21 Jul 2026 · InnerSource Commons Community Meetup
Leveraging AI in InnerSource
In this talk, I explored how AI can reduce friction throughout the InnerSource contribution journey. Using established InnerSource patterns, I showed how teams can accelerate contributions and collaborate more effectively.
Open Source & InnerSource
I contribute patterns and practices for making collaborative software development easier to scale across organizations.
August 2025 · InnerSource Common Patterns
AI Code Generation Context
AI-generated code can diverge from established project standards and architectural patterns. The AI Code Generation Context provides repository-specific guidance to help AI tools generate contributions that align with existing conventions, reducing review friction and improving code consistency.
July 2026 · InnerSource Common Patterns
Centralized Repository Governance
InnerSource repositories can lose contributor readiness as project hygiene drifts across teams. The Centralized InnerSource Repository Governance approach defines readiness checks as policy-as-code and uses automated audits to help Trusted Committers and maintainers keep repositories discoverable, understandable, and contribution-ready while preserving local ownership.
Articles
I write about the assumptions behind modern software engineering—where established patterns work, where they break, and the less obvious failure modes that emerge at scale.
My latest article
The most scalable API may be the one that does less
More expressive APIs give clients more freedom, but they also expose the underlying system to more unpredictable work. At scale, constraints can become a feature rather than a limitation.
ReadHTTP QUERY: The Standard Is Ready. Is Your System?
HTTP QUERY gives structured, read-only operations explicit semantics—but standardization doesn’t mean your system is ready. The real test is whether the entire request path preserves those semantics.
ReadDeployment Is Not CI/CD
No deployment process can eliminate failure. Good system design makes failure small, visible, and recoverable.
ReadWe Keep Talking About AI Context. But What Is Context?
AI keeps getting larger context windows. But context capacity is not context comprehension. Understanding how “context” changed across AI history reveals why that distinction matters.
ReadContext Poisoning: When AI Builds on the Wrong Truth
AI-assisted engineering turns documentation, comments, issues, prompts, and retrieved knowledge into inputs that can influence software decisions. What happens when those inputs are wrong?
ReadMore on AI Engineering · More on InnerSource · More on Software Engineering
Recognition
eBay Inc. · May 2026
Spot Award — AI Week
Samsung R&D Institute · April 2021
Excellent Awards Outstanding Project of the Year (R&D) — 2020-21
Samsung R&D Institute · April 2021
Best Community Engagement Award of the Year — 2020-21
Books that inspire me
A few books that have influenced how I think about software design, architecture, engineering organizations, and technical leadership.
