Principal Platform Engineer, Agentic AI Infrastructure

ESQlabs is an innovative, internationally active Contract Research Organization (CRO) and a global leader in the development and application of the Open Systems Pharmacology (OSP) Suite. We combine mechanistic modeling, Physiologically Based Pharmacokinetics (PBPK) and Quantitative Systems Pharmacology (QSP), with software and training to support Model-Informed Drug Development (MIDD) and Next-Generation Risk Assessment (NGRA) for pharmaceutical, chemical, and academic partners worldwide.

Our mission is to empower life sciences: helping scientists understand mechanisms of disease and chemical toxicity to optimize individualized, safer, and more efficient treatment and testing strategies.

We are now building an agentic AI framework on top of our open-source products: a platform in which AI agents orchestrate the OSP Suite and our other open-source tools to run regulated modeling workflows end to end.

You set up the agentic system that runs our analyses: the harness in which AI agents call our scientific tools under policy, the deterministic compute core they call, and the infrastructure that makes every run reproducible, metered, sealed and verifiable. The Quality Manager defines the standards; you implement them. You report to the CEO and work as a peer of our agentic engineering lead, with two platform engineers and the R developers who own the scientific tooling.

• Make the scientific compute core deterministic: pinned R, Python and solver tool chains, content-addressed container images with a software bill of materials, and tests that prove byte-identical re-execution on a reference platform and tolerance-bounded re-execution across platforms.

• Build the harness: tool router, policy engine (allow-lists, parameter bounds, autonomy levels), sandboxed per-tenant executors without egress, usage metering per tool call, and the signed provenance bundle and certificate issuer behind every result.

• Expose the platform as a versioned API and as tools for AI agents with one shared contract, and ship the reproducibility package that re-executes a run outside the platform.

• Run the evaluation pipeline in continuous integration: conformance tests against reference case sets, re-execution tests, regression tests, results stored per procedure version.

• Own release engineering and operations: CI/CD pipelines, signed and content-addressed releases, scripted installation qualification, environment management (hosted, in a client tenant, later offline), monitoring, incident handling and the operating model the deployment engineer runs.

• Implement the technical controls behind the information security and AI management standards the Quality Manager owns.

• At least five years of platform, infrastructure or backend engineering with senior ownership of a production system.

• Reproducible builds and execution in practice: dependency pinning, artifact hashing, immutable versioning, supply chain integrity. Hermetic build systems are a strong plus.

• Strong Python; comfort around R and scientific computing tool chains. Containers and infrastructure as code. Experience with workflow or DAG runtimes and with tool-calling systems for AI agents.

• API design where the schema was a contract with external consumers, including versioning and deprecation in practice.

• Security engineering for multi-tenant systems: isolation, secrets, authorization, and awareness of the risks of language-model-driven systems. Exposure to GxP or equivalent regulated software is valuable; interest in acquiring it is required.

• Dev Ops practice at production grade: CI/CD, Kubernetes or an equivalent orchestrator, observability, release management for environments you do not fully control.

The…