Agentic AI Engineer
First seen by Alion on Sep 23, 2026.
Key Responsibilities :
- Design and build LLM-powered agentic AI pipelines that reason, plan, and execute multi-step workflows with minimal human intervention.
- Own tool & function calling integrations connect LLMs to internal APIs, EHR systems, calendars, and third-party services.
- Build and maintain rigorous eval frameworks: offline eval sets, regression suites, latency benchmarks, and safety/guardrail tests.
- Ship and operate LLMs in production: streaming responses, structured outputs, prompt versioning, cost/latency tracking, and observability.
- Architect and maintain distributed microservices that expose LLM APIs async Python (FastAPI), event-driven pipelines, and real-time WebSocket workflows.
- Build RAG pipelines: chunking strategies, embeddings, vector stores, retrieval tuning, and PHI-safe handling.
- Take full ownership and conduct end-to-end testing for everything you build from unit and integration tests through to production validation and post-deployment monitoring.
- Collaborate closely with product, backend, and QA to define AI requirements and ship reliable, safe features.
Qualifications :
- 3+ years of hands-on Gen AI / LLM engineering experience building and shipping real systems, not just research or prototypes.
- 5+ years of hands-on AI / ML engineering experience overall (Gen AI + pre-Gen AI combined).
- Deep LLM expertise in prompt engineering, tool/function calling, structured outputs, chain-of-thought, and agent orchestration (LangChain, LangGraph, or similar).
- Proven eval culture that you treat evals as a first-class concern not an afterthought.
- Production LLM experience, integrations with OpenAI / Azure OpenAI / AWS Bedrock, streaming, cost optimisation, and monitoring.
- Strong software engineering fundamentals in Python (strong), async programming, REST APIs, CI/CD, Docker.
- Distributed systems fluency with microservice design, event-driven architecture (Kafka/SQS/RabbitMQ), and LLM API gateway patterns.
- End-to-end ownership mindset, you write tests, validate in staging, and don't consider a feature done until it's verified in production.
- Preferred SkillsNot required for hiring but these make you stand out:
- LLM post-training experience with fine-tuning, RLHF, DPO, or LoRA for domain-adapted or instruction-tuned models.
- Automatic prompt optimisation familiarity with tools or techniques like DSPy, TextGrad, or automated prompt search for systematic prompt improvement.
- Healthcare industry experience, working knowledge of EHR systems, HL7/FHIR standards, HIPAA compliance, or clinical workflows.
- Startup experience where you've worked in a fast-moving, resource constrained environment where you wore multiple hats and shipped quickly.
- Scaling & on-prem deployments experience deploying LLMs at scale, including self-hosted / on-prem model serving (vLLM, TGI, Triton) or hybrid cloud architectures.
Requirements added by the job poster :
- 5+ years of work experience with Applied Machine Learning
- 3+ years of work experience with Generative AI
- 3+ years of work experience with Python (Programming Language
Skills
Generative AI, Agentic AI, Python, EHR/EMR, FHIR, HL7, LLM, Healthcare EcoSystem, OpenAI, Azure, RAG