Solution Architect AI/ML Gen AI

Role & responsibilities

Position Dossier Solution Architect – AI/ML

About Us Omega Healthcare Management Services® (Omega Healthcare) is an AI-driven healthcare solutions company that partners across the healthcare ecosystem to deliver breakthrough results by reimagining and elevating revenue operations. Powered by the Omega Digital Platform®, our agentic AI engine leverages adaptive intelligence to drive automation, complemented by deep human expertise to help optimize performance and deliver sustained financial and clinical outcomes—while enhancing patient satisfaction. With a strong global presence spanning four countries, our team of 30,000+ dedicated professionals collaborate with hospitals, physician groups, and healthcare providers to streamline operations, improve financial performance and elevate patient care outcomes. We combine our deep domain expertise, advanced technology, and a customer-centric approach in delivering innovative and scalable healthcare support solutions.

Position Title: Solution Architect – AI/ML Department: Engineering & Technology Function: AI/ML Platform Engineering Role Summary The Data Science Solution Architect is a senior technical leadership role responsible for defining the enterprise architecture, scaling strategies, and production infrastructure for domain-specific SLMs, multi agent systems, and advanced retrieval architectures (including RAG and GraphRAG). This role bridges the gap between cutting-edge AI research and enterprise-grade deployment. The Architect owns the design of the foundational “agent harness”—setting the standards for secure tool execution, memory management, and guardrails—while ensuring models and retrieval pipelines are optimized for latency, cost, and accuracy across distributed hardware. Key Responsibilities • Architecture & Strategy: Define the enterprise architecture, standards, and roadmap for AI systems. Design models and retrieval pipelines optimized for accuracy, latency, and cost across diverse hardware envelopes (e.g., single-GPU, A100/H100 inference, and training envelopes). Establish the enterprise ontology and knowledge representation strategy. • GraphRAG & Advanced Retrieval Architecture: Lead the architectural design of hybrid retrieval systems. Scale Vector and Graph Databases (e.g., Neo4j) to handle massive enterprise corpora. Design systems that seamlessly route between standard semantic search and GraphRAG (community detection, multi-hop reasoning) based on query complexity. • Agent Harness Engineering & Orchestration: Architect the enterprise standard for multi-agent workflows (e.g., LangGraph, AutoGen). Design the core “harness”: robust state management, distributed memory architectures, secure tool-execution sandboxes, and deterministic guardrails (input/output validation, PII redaction, hallucination checks) that make autonomous agents safe for production. • Inference, Serving & Optimization: Own inference architecture using modern serving engines (vLLM, SGLang). Drive latency/throughput optimization via continuous batching, quantization, speculative decoding, FlashAttention, and optimized runtimes (TensorRT / TensorRT-LLM). • Distributed Training & Scaling: Lead multi-GPU training architectures (DDP / FSDP), implementing data- and model-parallel strategies, gradient accumulation, and Mixture-of Experts (MoE) scaling efficiency for specialized SLMs. • Cloud, MLOps & Production Readiness: Architect secure AI solutions on AWS or Azure. Define enterprise MLOps standards for CI/CD, adapter versioning, checkpointing, and CI/CD for agentic pipelines and knowledge graph updates. • Leadership & Mentorship: Act as the technical authority and design-review leader. Mentor Data Scientists and ML Engineers, fostering a research-driven culture that rapidly operationalizes the latest AI advancements. Required Skills & Expertise • Enterprise AI Architecture: Cloud-based ML architecture (AWS/Azure), MLOps, enterprise security, and AI governance. • Advanced Retrieval & GraphRAG: Deep architectural knowledge of Knowledge Graphs (Neo4j, Amazon Neptune), Vector Databases at scale, ontology design, and Microsoft GraphRAG or custom map-reduce graph reasoning patterns. • Agentic Frameworks & Harnessing: Multi-agent orchestration (LangGraph, AutoGen, CrewAI), enterprise harness engineering, secure tool routing, state persistence, and Guardrails AI/NeMo Guardrails. • Inference & Runtimes: Deep expertise in vLLM, SGLang, TensorRT, TensorR…