> Markdown version of [/jobs/ext/3853660-staff-ml-engineer-ai-platform-agents](https://www.wearedevelopers.com/jobs/ext/3853660-staff-ml-engineer-ai-platform-agents). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Staff ML Engineer AI Platform & Agents

- **Company:** ELLKAY, LLC.
- **Location:** United States (Remote available)
- **Experience:** Expert
- **Salary:** $180,000.0 - $220,000.0
- **Contract:** Permanent contract
- **Skills:** LangGraph Framework, Reference Implementation, Artificial Intelligence, Amazon Web Services, Automation of Tests, Data Infrastructure, Python (Programming Language), Machine Learning, Tensorflow, Azure Machine Learning, Software Engineering, SQL Databases, Feature Engineering, Pytorch, Fast Healthcare Interoperability Resources, Retrieval-Augmented Generation, Large Language Models, Snowflake, IT Architecture, Model Validation, Core Api, AI Platforms, Scikit Learn, Health Level Seven International, Machine Learning Operations, Api Design
- **Published:** October 21, 2026
- **Apply:** https://www.indeed.com/viewjob?jk=05f2a7eb26c4b124

## About the Role

* 8+ years of relevant software/ML engineering experience, with ownership of production ML systems and technical leadership across teams.
 * Strong Python and SQL, sound software design, automated testing, API development, and experience with PyTorch, scikit-learn, or comparable ML frameworks.
 * Hands-on model training, feature engineering, model selection, error analysis, statistical evaluation, and techniques to prevent dataset leakage.
 * Production experience with LLM applications, retrieval-augmented generation, embeddings, tool calling, model routing, and structured outputs.
 * Ability to design distributed services, cloud serving, access controls, model/prompt lifecycle management, and cost/performance tradeoffs.
 * Evidence of mentoring engineers and translating product needs into deliverable technical designs. A relevant degree or equivalent practical experience is acceptable., Healthcare integration or regulated enterprise AI; MCP/tool-calling experience; AWS and Snowflake; familiarity with HL7, FHIR, and migration or archive workflows. Framework experience such as LangGraph or an equivalent orchestration approach is useful.

## Description

Lead the design and delivery of shared AI/ML services for ELLKAY's migration, archive, connector, and interoperability products. This is a hands-on technical leadership role covering model development, LLM applications, governed agents, reusable serving interfaces, and engineering standards. The engineer will mentor the team and partner with Data Platform, product, clinical, and security leaders to turn pilots into measurable production capabilities., * Define the shared AI architecture, model-routing approach, retrieval interfaces, tool contracts, and release criteria. Lead design reviews and mentor the distributed ML team.

 * Design governed agent experiences for implementation guidance, migration recommendations, operational explanations, and archive retrieval.
 * Integrate with the shared agent gateway and existing platform APIs. Enforce tenant and purpose checks, approvals for actions, async job status, and auditable tool execution.
 * Choose between rules, classical ML, retrieval, and LLMs using measured quality, latency, and cost. Coordinate rollout from synthetic sandbox to approved customer pilots.
 * Develop reference implementations and own difficult production failures, including model regressions, retrieval errors, invalid tool actions, and latency/cost overruns.Qualifications and expected outcomes, * First 90 days: agree the AI architecture and ownership boundaries, establish evaluation/release standards, and deliver a reference implementation for the first two use cases.
 * By six months: deliver a common AI service consumed by two workflows, mentor the four specialists, and complete a synthetic or no-PHI agent pilot with measured task quality, latency, and unit cost.
 * By 12 months: demonstrate common-service adoption by at least two product families, a measured improvement in an agreed customer workflow, and reproducible evaluation evidence for 100% of production AI releases.
 * Quality, performance, and business-impact targets will be calibrated to the selected use cases and their initial baselines. Delivery depends on the corresponding platform capabilities and approved customer pilot scope.