Earning Agentic Autonomy: An Eval Architecture for AI Agents
How evaluation evidence, explicit controls, and runtime enforcement can help determine when an AI agent is ready to act with greater autonomy.
AI ENGINEERING · PLATFORM ENGINEERING · RELIABILITY
Engineering AI for the real world.
I’m Ravikumar Buragapu, a Director of Engineering focused on AI engineering, platform engineering, and reliability at enterprise scale. I work on turning emerging technologies—from GenAI to agentic systems—into platforms that are secure, observable, resilient, and ready for production.
My perspective is simple: the next phase of AI will be shaped not just by what models can do, but by how we engineer the systems around them—identity, evaluation, permissions, infrastructure, operational controls, and human accountability.
Explore engineering·Read my writing·Speaking
Explore BenchmarkKit, an open-source LLM and RAG benchmarking toolkit·Explore Personal Chief of Staff, a multi-agent MCP assistant·Explore Chat with My Research, a citation-first RAG project·Explore Learned ANN Index, a vector-search research prototype·Explore Tokenizer Ablation, a from-scratch BPE and downstream study·Explore RNN vs Transformer, a parameter-matched sequence study·Explore Chunking Ablation, a controlled RAG retrieval study·Explore the Validated RAG Eval Harness, a human-vs-judge agreement study
What I focus on
AI and agentic systems
Designing the engineering foundations for production AI: agent identity, least-privilege access, evaluation, guardrails, observability, and controlled autonomy.
Platform engineering and developer productivity
Building platforms, reusable engineering capabilities, and automation that help teams deliver software faster while maintaining security, consistency, and quality.
Reliability and operational intelligence
Engineering resilient, data-intensive systems through observability, SRE practices, incident automation, capacity engineering, and measurable service objectives.
Engineering leadership at scale
At Oracle Health, I lead engineering across infrastructure, reliability, and data intelligence platforms serving commercial, regulated, and Federal environments.
Previously, at Wells Fargo, I formed the Agentic and GenAI Squad and helped shape agentic engineering strategy across an organization reaching more than 10,000 developers. At Adobe, I led platform engineering, the Agentic AI Builders Group, and development of an AI-driven Service Maturity Insights platform.
My experience also includes engineering roles at Kohl’s and Charles Schwab, building on a software engineering career that began in India in 2000.
Ideas on engineering what comes next
I write about the engineering decisions that determine whether AI systems can be trusted in production—from evaluation and identity to governance, operational reliability, and the boundaries of agent autonomy.
Let's exchange ideas
I’m interested in conversations about production AI, agentic architectures, platform engineering, reliability, and the future of engineering organizations.