Experimenter Jobs (NOW HIRING)

We are seeking a Staff Machine Learning Engineer with deep expertise in agentic AI — and a true passion for experimentation and creation — to design, build, test, evaluate, and productionize next-generation autonomous AI agents for healthcare and clinical research. If you love rapidly prototyping wild ideas, running build-test-learn cycles, iterating on novel agent behaviors, and turning unsolved challenges into working systems, this is the role for you. You will own end-to-end agentic workflows that reason, plan, use tools, orchestrate multi-agent collaboration, and deliver safe, reliable outcomes in highly regulated environments, while collaborating with multidisciplinary teams to influence Medeloop's technological direction. You will also be nested within a team of advisors and collaborators with deep medical and health expertise, including scientists, clinicians, and AI experts, including the former FDA commissioner, former editor of JAMA, and developer of BloombergGPT. The result: You will be an active participant in fostering a data lead public health and healthcare ecosystem.

• Lead the design and architecture of advanced agentic AI systems, including reasoning loops (ReAct, CoT, ToT), tool-calling, dynamic multi-agent orchestration, RAG pipelines, memory/state management, and emerging protocols like Model Context Protocol (MCP) and Agent-to-Agent (A2A).
• Build and own production-grade agent infrastructure, including prompts, function tools, workflow graphs, MCP/A2A integrations, and adaptive agent lifecycle management (spinning up, specializing, delegating, and decommissioning agents dynamically for complex healthcare workflows).
• Develop rigorous evaluation and safety frameworks — automated testing, benchmarking, regression testing, adversarial testing, safety guardrails, observability (tracing, logging, metrics), and human-in-the-loop mechanisms to ensure reliable, compliant performance in production.
• Drive LLM and ML model development — train, fine-tune, and deploy large-scale models on healthcare datasets, working closely with researchers and clinicians to solve real clinical challenges.
• Shape Medeloop's agentic AI strategy and roadmap in close partnership with the C-suite and cross-functional leadership.
• Stay at the cutting edge of agentic AI (multi-modal agents, advanced reasoning models, interoperability protocols) and help establish Medeloop as a leader in transparent, compliant healthcare AI.

• 7+ years of hands-on experience as a Machine Learning Engineer, with a proven track record building and shipping production agentic AI systems (single- or multi-agent) in industry, ideally in healthcare, life sciences, or other related domains.
• Experience working on analytic engines (or advanced analytics platforms) — designing, optimizing, or integrating systems that power data-driven insights, queries, or decision-making at scale.
• Strong theoretical foundation in ML/AI, with emphasis on NLP/LLMs, reinforcement learning, planning/reasoning algorithms.
• Deep expertise with agentic frameworks and tools: LangChain/LangGraph, Model Context Protocol (MCP), Agent-to-Agent (A2A) protocols, Hugging Face, PyTorch, vector databases/semantic search, prompt engineering, and observability platforms (e.g., LangSmith, Phoenix).
• Experience designing fully automated evaluation and testing pipelines for autonomous agents and their orchestration, including metrics for reliability, safety, factuality, cost/latency, clinical utility, and dynamic behaviors.
• A builder/experimenter mindset — you thrive on rapid prototyping, testing bold new ideas, iterating quickly on agent designs, and exploring uncharted territory in agentic systems.
• Passion for unsolved challenges in healthcare AI, with the ability to thrive in a fast-paced, multidisciplinary environment and wear multiple hats.

• Strong record in top AI/ML conferences/journals; experience with healthcare data (EHRs, claims) and regulatory considerations (HIPAA, transparency, reproducibility).
• Multi-cloud experience (AWS, Azure, GCP)

• Ownership from day one: small team, high-trust, no layers between your work and its impact
• Technically ambitious: you'll build AI-powered workflows, not just support them
• Real-world stakes: your work accelerates drug development, addresses health equity, and improves clinical research for institutions that matter
• Strong foundation: Series A, top-tier investors, and a data asset (200M+ patient records) that most companies spend years trying to build