Runtimes and harnesses
I design typed tools, provider-neutral boundaries, context and retrieval flows, evaluation loops, and observability so teams can reason about agent behavior instead of guessing.
AI engineering / systems / product
I'm Indra Zulfi Mushoddaq—an AI Engineering Lead, systems builder, and founder based in Indonesia.
I work where agent behavior, backend architecture, and product experience meet. I turn high-context AI ideas into systems that are testable, observable, repeatable, and useful in production.
What I bring
The work is strongest when these capabilities stay connected: the agent runtime, the product around it, and the engineering practice required to operate both.
I design typed tools, provider-neutral boundaries, context and retrieval flows, evaluation loops, and observability so teams can reason about agent behavior instead of guessing.
I move across backend execution, data, streaming interfaces, and deployment—useful when an AI feature needs one owner from prototype through production.
I lead engineering, build independently when needed, and create reusable tools and education that make other engineers faster and the product easier to operate.
Selected evidence
Created an open-source provider-neutral runtime boundary for AI applications.
Trained 600+ students across 8 programs and 24 cohort batches.
Helped power 88K+ research reports and 45K+ analyses across AlphaX research workflows.
Owned product flow, backend behavior, tool workflows, analysis execution, and AI-assisted reporting.
Combined chat, tasks, terminal, tools, and git context into one workspace.
Exposed code search, symbol lookup, outlines, and dependency tracing.
Relevant experience
AI Engineering Lead
Japan · Remote · Aug 2024 - Present
Founder and Mentor
Indonesia · Hybrid · May 2023 - Present
Tech Generalist and Software Engineer
Canada · Remote · Aug 2020 - May 2024
Engineering notes
How Codex, GPT-5.6 Sol, Herdr, Zed, Git worktrees, and bounded feedback loops fit into my day-to-day engineering workflow.
I am building Anvia because I care about lower-level AI runtime primitives, explicit TypeScript abstractions, and developer experience.
A practical case for SQLite: individual queries get much cheaper when there is no network in the middle.