Key Responsibilities:

- End-to-End Lifecycle Ownership (Build, Deploy, Maintain & Improve)

1. Build: Design and code autonomous AI agents, custom API tools, and Retrieval-Augmented Generation (RAG) pipelines using modern orchestration frameworks (LangGraph, CrewAI, AutoGen, etc).
2. Deploy: Set up robust deployment pipelines to move AI agents from local MVPs to production-grade environments using Docker, APIs, microservices, and cloud infrastructure.
3. Maintain: Take full responsibility for the operational health of your deployed agents - monitor uptime, handle error states, manage token/cost limits, and ensure day-to-day reliability.
4. Improve: Establish systematic evaluation loops (Evals) to measure agent accuracy. Continuously refine prompt structures, system guidelines, guardrails, and memory retrieval mechanisms to improve agent performance over time.

- Technical Implementation & Data Foundation

1. Agentic Frameworks: Build and manage autonomous AI agents for tasks such as lead prospecting, market segmentation, content generation, claims/coding workflows, operations management, and data synthesis.
2. Data Integration: Implement and optimize indexing, hybrid search, and RAG pipelines using vector databases (Pinecone, Qdrant, Milvus, Chroma) or knowledge graphs (Neo4j) to supply high-context data to agents - critical for RAAPID’s grounded use cases.
3. API Integration & Tooling: Connect LLMs to internal enterprise systems (CRM, EHR/FHIR endpoints, databases, operational tools) and external APIs, enabling agents to execute real-world actions safely.

- AI Reasoning,



Evaluations & Guardrails

1. Reasoning & Evaluation: Define, test, and run automated evaluation frameworks (Ragas, LangSmith, or custom eval datasets) to track latency, accuracy, safety, and hallucination rates across agent workflows.
2. Guardrails & Security: Implement strict guardrails to ensure agents operate safely, protect PHI/PII, and adhere to HIPAA, HITRUST, and other compliance standards intrinsic to healthcare AI.
3. Cost & Latency Optimization: Tune model selection, caching, batching, and prompt design to keep token economics and end-to-end latency within production SLOs.

- Cross-Functional Collaboration

1. Collaborate with Translators: Partner with CoE Translators to convert business workflows into production-ready agent architectures, treating their domain context as a first-class design input.
2. Support Business Units: Act as a technical advisor to Sales, Marketing, Product, Operations, HR, Legal, and Finance Champions on feasibility, ROI, and best practices for automating their workflows through Agentic AI.
3. Knowledge Sharing: Document architectural decisions, share reusable agent patterns, and contribute to the CoE’s internal library of components, prompts, evals,



and guardrails.

Preferred Qualifications

Agentic Frameworks: Hands-on experience with multi-agent orchestration
libraries (CrewAI, AutoGen, LangGraph, etc).
Data & RAG Systems: Experience with Vector Databases (Pinecone, Qdrant, Milvus,
Chroma) and advanced retrieval techniques (hybrid search, reranking, query
rewriting, GraphRAG).
LLM Evaluation & Ops: Experience building LLM evaluation suites (Evals) and
monitoring agents in production using LangSmith, Ragas, or comparable tooling.
Cloud & Deployment: Familiarity with deploying applications to cloud
environments (Azure preferred; AWS or GCP acceptable) using containerization
(Docker, Kubernetes).
Healthcare/Regulated Domain Exposure: Prior work in healthcare, fintech, or other
regulated industries - is a robust plus.
Specialization Tracks: We are particularly interested in candidates who bring
depth in one of three areas -

1. AI Reasoning / LLM expertise & Evals,
2. Data Foundation (indexing, retrieval, knowledge graphs)
3. end-to-end Agent Engineering and tool integration

Technical Skills
Languages: Python (highly preferred), SQL, Node.js / TypeScript.
AI Frameworks: LangChain, LlamaIndex, AutoGen, CrewAI, LangGraph, Semantic
Kernel.
Databases: PostgreSQL, MongoDB, Pinecone, Qdrant, Neo4j.
Tools: Git, Docker, FastAPI / FastStream, LangSmith, Ragas (for tracing and evals).
Cloud: Azure (preferred), AWS, or GCP - with containerization and CI/CD
experience.

📌 Senior AI Engineer (Gujarat)
🏢 TALCHEMY SOLUTIONS
📍 Gujarat