Calvin
Mark
Calvin
Mark
Applied intelligence.
Agentic platforms.
Product.
Human-centered design.
Applied intelligence.
Agentic platforms.
Product.
Human-centered design.
Applied intelligence.
Agentic platforms.
Product.
Human-centered design.
I design and build AI systems that make complex intelligence feel simple.
Experience
2024 – Now
AI Engineer
Github Copilot
Developed an asynchronous, cloud-based coding agent focused on session management, memory primitives, and reliability in production.
Improved runtime stability and observability across distributed systems, and built a multi-OS execution stack for self-hosted environments.
Collaborated across product and platform teams to evaluate system tradeoffs and deliver trustworthy AI experiences.
2024 – Now
AI Engineer
Github Copilot
Developed an asynchronous, cloud-based coding agent focused on session management, memory primitives, and reliability in production.
Improved runtime stability and observability across distributed systems, and built a multi-OS execution stack for self-hosted environments.
Collaborated across product and platform teams to evaluate system tradeoffs and deliver trustworthy AI experiences.
2024 – Now
AI Engineer
Github Copilot
Developed an asynchronous, cloud-based coding agent focused on session management, memory primitives, and reliability in production.
Improved runtime stability and observability across distributed systems, and built a multi-OS execution stack for self-hosted environments.
Collaborated across product and platform teams to evaluate system tradeoffs and deliver trustworthy AI experiences.
2020-2024
Software Engineer
Oracle
Contributed to internal AI platform tooling across multiple domains, including computer vision, code indexing, and internal coding agents.
Developed AI systems deployed on OCI, supporting production workloads and government-facing DevOps environments.
Contributed to internal AI safety and reliability tooling, and partnered across teams to improve usability, performance, and operational robustness of AI services at scale.
2020-2024
Software Engineer
Oracle
Contributed to internal AI platform tooling across multiple domains, including computer vision, code indexing, and internal coding agents.
Developed AI systems deployed on OCI, supporting production workloads and government-facing DevOps environments.
Contributed to internal AI safety and reliability tooling, and partnered across teams to improve usability, performance, and operational robustness of AI services at scale.
2020-2024
Software Engineer
Oracle
Contributed to internal AI platform tooling across multiple domains, including computer vision, code indexing, and internal coding agents.
Developed AI systems deployed on OCI, supporting production workloads and government-facing DevOps environments.
Contributed to internal AI safety and reliability tooling, and partnered across teams to improve usability, performance, and operational robustness of AI services at scale.
Thesis
I believe the best use of intelligence is mostly invisible.
(the kind you barely realize is there….)
Thoughtfully designed systems hide complexity by default and surface it only when needed, allowing users to focus on outcomes rather than the system itself.
Creative systems master the discipline of making complex intelligence feel simple, guided by restraint, clarity, and intention rather than novelty for its own sake.
Intelligent systems must be trustworthy, auditable, reliable, and safe by design. I treat these qualities not as constraints, but as creative inputs and a reflection of my design judgment.
I believe the best use of intelligence is mostly invisible.
(the kind you barely realize is there….)
Thoughtfully designed systems hide complexity by default and surface it only when needed, allowing users to focus on outcomes rather than the system itself.
Creative systems master the discipline of making complex intelligence feel simple, guided by restraint, clarity, and intention rather than novelty for its own sake.
Intelligent systems must be trustworthy, auditable, reliable, and safe by design. I treat these qualities not as constraints, but as creative inputs and a reflection of my design judgment.
The best use of intelligence is mostly invisible.
(the kind you barely realize is there….)
Thoughtfully designed systems hide complexity by default and surface it only when needed, allowing users to focus on outcomes rather than the system itself.
Creative systems master the discipline of making complex intelligence feel simple, guided by restraint, clarity, and intention rather than novelty for its own sake.
Intelligent systems must be trustworthy, auditable, reliable, and safe by design. I treat these qualities not as constraints, but as creative inputs and a reflection of my design judgment.
Active Independent Work
Agentic Systems
(Confidential)
Exploration of long-running AI systems that operate within real organizational contexts. The work focuses on agent behavior over time, decision support, human-in-the-loop design, and reliability under real-world constraints such as trust, safety, and adoption.
Memory & Context Systems
(Confidential)
Foundational work on memory and context architectures for intelligent systems, focused on how information persists, decays, and influences behavior across extended workflows. This exploration informs how agentic systems maintain continuity, judgment, and alignment over time.