Senior Agentic AI Engineer — Agentic Medical AI at SAIGroup

<div id="JobDetails_AboutTheJob_4458493249" class="f8b0bb6c _4f72659f _952d2cfe"> <div data-sdui-component="com.linkedin.sdui.generated.jobseeker.dsl.impl.aboutTheJob"> <div class="_7afdf9b8 c313cecd d52ea82d _3e244506 dacc36d3 _832ba387 _1d289483 _952d2cfe"> <div class="_7afdf9b8 _7a2fcf01 fd6601ce fa4e965b da48b274 _4eda3a6d ea670751 _70f99203 d71d767a c313cecd d52ea82d _3e244506 dacc36d3 _832ba387 _1d289483 f6e98cd8 _0e7ab03d a88af83b _854860ea fa63ca3b _952d2cfe"> <div class="_7afdf9b8 _6dc3867d a54932dd _060d2dd3 _1903fd51 c313cecd d52ea82d _3e244506 dacc36d3 _832ba387 _1d289483 _952d2cfe"> <p class="e7c0a629 _3e5ea668 cb45d839 _08acb8d4 _900db78a _836e8169 _1e296b88 _24c6998a _8dbc9ff2 _6a5cfddc d4eddadc"><span class="d07f5a6d ecbf71c2 _509dbddf _952d2cfe" data-testid="expandable-text-box"><strong>About the Role<br><br></strong>We are building a new class of <strong>clinical-grade AI agents</strong> powered by a foundational model that unifies <strong>3D volumetric medical data</strong> and <strong>textual reasoning</strong>. These agents will assist with radiology workflows, treatment planning, case summarization, longitudinal patient analysis, and human-in-the-loop clinical decision support.<br><br>In this role, you will design the <strong>agent layer</strong> that transforms a powerful multimodal foundation model into a safe, reliable, adaptive medical assistant capable of analyzing 3D scans, retrieving past cases, explaining findings, and coordinating multi-step reasoning chains.<br><br>This position is ideal for researchers from multimodal/VLM/CLIP ecosystems who want to move up the stack into <strong>autonomous medical reasoning</strong>, multimodal alignment, and real-world clinical impact.<br><br><strong>What You Will Work On<br><br></strong></span></p> <ul class="_9e130f2e d4b40f94 _952d2cfe"> <li>Architect LLM-driven medical agents capable of:</li> <ul class="_9e130f2e d4b40f94 _952d2cfe"> <li>Reading &amp; interpreting 3D medical volumes (MRI, CT, PET).</li> <li>Answering radiology questions, generating report drafts, and performing differential reasoning.</li> <li>Retrieving similar prior cases using contrastive embeddings.</li> <li>Synthesizing multi-instance evidence across scans, labs, and notes.</li> <li>Executing multi-step “clinical workflows”—image → analysis → summary → recommendation.</li> </ul> <li>Integrate clinical agents with the 3D foundation model’s representations, enabling structured medical reasoningand contextual memory across longitudinal data.</li> <li>Develop evaluation frameworks for trustworthiness, interpretability, calibration, and clinical safety.</li> <li>Collaborate with clinicians and scientists to co-design tasks, datasets, and validation protocols.</li> <li>Benchmark models against open and proprietary datasets: MIMIC, CheXpert, MosMed, LiTS, BraTS, and India-specific clinical corpora.</li> <li>Publish findings at top-tier venues (MICCAI, NeurIPS, AAAI, EMNLP).</li> <li>Work with regulatory and product teams to translate research into medically viable systems.</li> </ul> <strong>Why This Role Appeals to CLIP / VLM / Chitrarth / Patram Authors<br><br></strong> <ul class="_9e130f2e d4b40f94 _952d2cfe"> <li>A rare opportunity to work at the intersection of agentic LLMs, multimodal 3D vision, and clinical intelligence.</li> <li>Ability to bring contrastive learning, cross-lingual modeling, multimodal alignment, and document understanding into the medical domain.</li> <li>Greenfield space: medical agent design is still unexplored compared to 2D VLMs.</li> <li>Freedom to publish, open source, and propose new clinical reasoning benchmarks.</li> <li>Attractive to researchers who want impact beyond academic settings—directly improving physician workflows and patient outcomes.</li> <li>A chance to create India’s first medical foundation-model agent system, setting global standards.<br><br></li> </ul> <strong>What We’re Looking For<br><br></strong> <ul class="_9e130f2e d4b40f94 _952d2cfe"> <li>Experience developing LLM-based agents, tool-using models, chain-of-thought systems, or orchestrated pipelines.</li> <li>Strong background in multimodal learning, vision–language transformers, contrastive models, or clinical NLP.</li> <li>Familiarity with medical imaging, radiology workflows, DICOM pipelines, or 3D data representations.</li> <li>Expertise in PyTorch/JAX and large-scale training and inference.</li> <li>Ability to design safe, interpretable, and auditable reasoning systems.</li> <li>Strong publication record in multimodal, medical AI, agentic systems, or LLMs.</li> <li>Comfort collaborating with clinicians and biomedical researchers.<br><br></li> </ul> <strong>Nice to Have<br><br></strong> <ul class="_9e130f2e d4b40f94 _952d2cfe"> <li>Experience with agent frameworks (LangChain, LLaMA Index, Haystack, custom pipelines).</li> <li>Background in clinical ontologies, SNOMED, RadLex, or UMLS.</li> <li>Contributions to medical AI challenges (BraTS, RSNA, MedVQA, MedIC).</li> <li>Prior work in multilingual or code-mixed medical text.</li> <li>Proven ability to take research concepts into production-grade systems.<br><br></li> </ul> <strong>What We Offer<br><br></strong> <ul class="_9e130f2e d4b40f94 _952d2cfe"> <li>Competitive compensation.</li> <li>Access to high-compute clusters + large proprietary datasets.</li> <li>Full freedom to publish and define development direction.</li> <li>Cross-functional collaboration with clinicians, regulatory experts, and product teams.</li> <li>A clear path to building systems that meaningfully improve healthcare outcomes.</li> </ul> <p>&nbsp;</p> </div> </div> </div> </div> </div> <div id="JobDetails_JobAlertToggle_4458493249" class="f8b0bb6c _4f72659f _952d2cfe"> <div data-sdui-component="com.linkedin.sdui.generated.jobseeker.dsl.impl.jobAlertToggle"> <div class="_7afdf9b8 c313cecd d52ea82d _3e244506 dacc36d3 _832ba387 _1d289483 _952d2cfe"> <div class="_7afdf9b8 _7a2fcf01 fd6601ce fa4e965b da48b274 _4eda3a6d ea670751 _70f99203 d71d767a c313cecd d52ea82d _3e244506 dacc36d3 _832ba387 _1d289483 f6e98cd8 _0e7ab03d a88af83b _854860ea fa63ca3b _952d2cfe"> <div class="_7afdf9b8 _6dc3867d a54932dd _060d2dd3 _1903fd51 c313cecd d52ea82d _3e244506 dacc36d3 _832ba387 _1d289483 _952d2cfe"> <div class="c313cecd _725682b3 _6ef60798 dacc36d3 _832ba387 _1d289483 _952d2cfe">&nbsp;</div> </div> </div> </div> </div> </div> <p>&nbsp;</p>