ML/AI Engineer with Platform expertise — building durable infrastructure for ML and agent systems
Core maintainer at ZenML for four years, working at the intersection of Machine Learning, platform engineering and lately AI Agents. My days move between Kubernetes, observability, and durable execution — and lately, a lot of agent infrastructure.
Now — hardening Seshat's deep research engine (100+ checkpoints per run, 14 agents, Logfire observability).
- ZenML — core maintainer. MLOps orchestration for production ML and agent pipelines.
- Kitaru — durable execution SDK for agentic workflows Built on ZenML.
- Seshat — multi-agent deep research reading platform. Checkpointed pipelines, Chain-of-Verification, freshness-first retrieval.
Languages Python · Go · Bash
AI PydanticAI · Claude Code · MCP · vLLM · SGLang · Triton
Platform Kubernetes · EKS · Terraform · KServe · Envoy Gateway · OpenTelemetry
I publish at safoine.com/blog — notes on agent infrastructure, durable execution, MLOps at scale, and build-in-public retros. Recent threads usually surface on LinkedIn and X first.
Open to chats on MLOps, agent infra, durable execution, or Kubernetes in general. Reach out at safoinme@gmail.com or safoine.ext@zenml.io.
Tidy laptop, messy desktop — still my slogan.




