I'm building AI benchmarking and alignment infrastructure for Bespoke Labs through a.team, designing Kubernetes and Docker environments for evaluating frontier AI agents in large-scale SRE scenarios.
I've architected adversarial failure scenarios with Istio, ArgoCD, Jaeger, Keycloak, OpenTelemetry, HMAC-SHA256 security gates, and Python grading engines for end-to-end infrastructure-as-code validation.
At Turing, I built Python/FastAPI backends, AI workflows, agentic systems, RAG patterns, and REST integrations between LLM providers and React interfaces. I also standardized CI/CD and testing to support reliable production AI features.
Earlier, at ProximusDA and Instigate Design, I optimized FPGA and GPU applications with OpenCL, built embedded C++ components that improved performance by 30%, and developed a semantic search solution for encrypted data.
