Research & Engineering
Building reliable infrastructure for AI agents - from prototype to production.
What I'm building
LLM agents need infrastructure that is both efficient (handling long contexts, optimizing memory, reducing redundancy) and secure (enforcing least-privilege access, resisting adversarial manipulation). I build systems at this intersection - from research prototypes to shipped products. Memory management, context optimization, authorization, and observability.
Focus areas
Context Efficiency & Reliability
"How do we make LLM outputs reliable and deterministic through better context management?"
Building systems that clean, deduplicate, and optimize context before it reaches the model. Deterministic algorithms over probabilistic heuristics.
Agent Authorization & Audit Trails
"How do we enforce fine-grained, capability-based authorization for AI agents with full auditability?"
Applying Google Zanzibar-style authorization models to agent-tool interactions via OpenFGA and MCP. Dynamic capability tokens, real-time policy enforcement, and audit logging.
Adversarial Robustness & Observability
"How can agents maintain safety under prompt injection, tool poisoning, and adversarial tool responses?"
Building observability and tracing infrastructure to detect and mitigate attacks on agent tool-use pipelines. GPU-level profiling for performance and security.
Products & Prototypes
Distill
Deterministic context deduplication for LLMs in ~12ms
Agent Authorizationagentic-authz
OpenFGA + MCP authorization gateway for AI agents
ObservabilityLLMTraceFX
GPU-level LLM inference profiler with AI-powered insights
AuthorizationOpenFGA
Google Zanzibar-style authorization (Core Maintainer)
Context EngineeringContextLab
LLM context engineering toolkit - analyze, compress, visualize
Memory SystemsTokenVM
KV-cache virtual memory with page-based eviction
Agent SystemsAgentflow
Agent orchestration with capability-based security
SecuritySentinel AI
Security scanning with LLM-powered triage
SecurityA2AS
Agent-to-agent security protocol implementation
DevSecOpsactionsec
GitHub Actions security analyzer
Working with me
I bring a combination of systems engineering experience and research curiosity. Here's what I offer as a collaborator:
- Shipping: from research prototype to production - built Distill, contribute to OpenFGA as maintainer
- Systems depth: production infrastructure at Ona (formerly Gitpod)
- Open source: OpenFGA maintainer (CNCF), GitHub1s maintainer
- Cross-domain: security, distributed systems, ML infrastructure
- Communication: technical writing on Dev.to and Medium