AI/ML Engineer

New York Technology Partners; United States; On-site; USD 180,000 - 240,000; Full time

Job summary

New York Technology Partners is seeking a senior AI systems engineer to design, deploy, and operate LLM-powered applications with agent-based orchestration and robust RAG pipelines. You will collaborate with clinical and product teams to translate complex problems into reliable AI solutions.

The role requires hands-on experience with LLMs, multi-agent frameworks, and end-to-end ML pipelines, plus strong software engineering in Python and cloud-native architectures.

Qualifications

• Strong grounding in ML fundamentals and evaluation methodologies.
• Hands-on with deep learning frameworks (PyTorch or TensorFlow).
• Solid NLP knowledge and transformer architectures.

Responsibilities

• Design, build, and deploy LLM-powered, agentic AI applications with multi-agent orchestration.
• Develop and optimize RAG pipelines with provenance and factual grounding.
• Build document intelligence workflows for extraction, classification, OCR, and structuring.

Skills

ML fundamentals

Deep learning

NLP

LLMs

Generative AI

RAG

Vector databases

Knowledge graphs

AI agents

Agentic AI frameworks

Python

Java

APIs

Microservices

Distributed systems

Azure AI Services

AWS SageMaker

Google Vertex AI

Databricks

Snowflake

Spark

Kafka

Kubernetes

Docker

GitHub Actions

Azure DevOps

Terraform

CI/CD

Responsible AI

Education

Bachelor's or Master's in Computer Science/Data Science/ML

Tools

LangChain

LangGraph

LlamaIndex

OpenAI

Azure OpenAI

Spark

Airflow

Job description

• Strong expertise in Machine Learning, Deep Learning, NLP, Large Language Models (LLMs), and Generative AI.
• Hands-on experience with RAG, Vector Databases, Knowledge Graphs, AI Agents, and Agentic AI frameworks.
• Proficiency in Python, Java, APIs, Microservices, and distributed system architecture.
• Strong knowledge of Azure AI Services, Azure OpenAI, AWS SageMaker, or Google Vertex AI.
• Experience with Databricks, Snowflake, Spark, Kafka, and modern data engineering platforms.
• Expertise in Kubernetes, Docker, GitHub Actions/Azure DevOps, Terraform, and cloud-native architectures.
• Knowledge of Responsible AI, AI Governance, Model Risk Management, and AI Security principles.

Key Responsibilities

• Design, build, and deploy LLM powered and agentic AI applications, including multi agent orchestration, tool/function calling, and MCP based integrations.
• Develop and optimize RAG pipelines — chunking, embedding, retrieval, re ranking, and grounding — with a focus on provenance and factual accuracy.
• Build document intelligence workflows (extraction, classification, OCR, structuring) over complex clinical and operational documents.
• Implement human in the loop review gates and feedback loops that let subject matter experts correct, validate, and improve model output.
• Instrument systems for ML observability — latency, cost, token usage, drift, and quality — and act on what the telemetry shows.
• Write production grade Python, containerize services, and operate them through CI/CD and orchestration tooling.
• Collaborate with product, clinical, and operations partners to translate ambiguous business problems into reliable AI systems.
• Uphold Responsible AI practices: evaluation, bias/error analysis, guardrails, and clear documentation.

Required Skills

• Strong grounding in ML fundamentals — supervised/unsupervised learning, evaluation methodology, and model selection.
• Practical experience with deep learning frameworks (PyTorch and/or TensorFlow).
• Solid understanding of NLP and transformer architectures.

Generative AI, LLMs & Agentic Systems

• Hands on experience building with LLMs (OpenAI/Azure OpenAI, Anthropic Claude, or comparable).
• Prompt engineering, structured outputs, and function/tool calling.
• Experience with agentic frameworks and orchestration (e.g., LangChain, LangGraph, LlamaIndex, or equivalent) and multi agent design patterns.
• RAG system design: vector databases, embeddings, retrieval and re ranking strategies, and grounding/citation techniques.
• Familiarity with the Model Context Protocol (MCP) or similar tool/integration standards.
• Strong SQL and experience with relational databases (SQL Server, PostgreSQL, or similar).
• Building and maintaining data/ML pipelines and workflow orchestration (Airflow or equivalent).
• Comfort working with unstructured and semi structured data at scale.

MLOps & Observability

• Model/LLM evaluation frameworks and offline/online testing.
• Observability tooling for AI systems (e.g., Arize, Langfuse, or comparable) — monitoring quality, cost, drift, and token usage.
• Experiment tracking and reproducibility practices.

Software Engineering & Cloud

• Expert level Python and sound software engineering habits (testing, code review, version control with Git/GitHub).
• Containerization with Docker and CI/CD (GitHub Actions or equivalent).
• Cloud platform experience (Azure preferred; AWS/GCP acceptable), including deploying and scaling services.

Preferred / Nice to Have

• Experience with AI in healthcare, insurance, or another regulated, high-stakes domain.
• Familiarity with Responsible AI frameworks, model governance, and validation gates.
• Experience designing UIs or interaction patterns for AI assisted expert review (trust, explainability, click to evidence).
• Kubernetes and infrastructure as code exposure.
• Experience working with clinical or operational subject matter experts.

Qualifications

• Bachelor's or Master's in Computer Science, Data Science, Machine Learning, or a related field (or equivalent practical experience).
• [5]+ years building and shipping ML/AI systems, including recent hands-on work with LLMs or agentic applications.

What Makes You a Fit

• You care about correctness and are comfortable engineering for a world where AI errors will happen — building the guardrails, evaluation, and human oversight to catch them.
• You can communicate technical trade offs to non-technical clinical and business partners.
• You take ownership, work deliberately, and iterate based on real usage and data.