> Markdown version of [/jobs/ext/4505265-senior-ai-ml-engineer](https://www.wearedevelopers.com/jobs/ext/4505265-senior-ai-ml-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Senior AI/ML Engineer

- **Company:** Sumeru Solutions
- **Location:** United States
- **Experience:** Expert
- **Contract:** Permanent contract
- **Skills:** LangGraph Framework, Application Programming Interfaces (APIs), Artificial Intelligence, Cloud Computing, Continuous Integration, Graph Database, Identity and Access Management, Python (Programming Language), Routing, Role-Based Access Control, Software Engineering, Enterprise Data Management, Datadog, Data Logging, Scripting, LangChain, Evaluation Pipelines, Retrieval-Augmented Generation, Large Language Models, Snowflake, Multi-Agent Systems, Prompt Engineering, Llamaindex, Caching, Backend, Agentic-AI, Fastapi, Event Driven Architecture, Integration Tests, Low Latency, Apache Kafka, Evaluation of Large Language Models, Invoking Functions, Restful APIs, Model Context Protocol, Amazon Simple Queue Service (SQS), Agent2Agent Protocol
- **Published:** November 2, 2026
- **Apply:** https://www.careerbuilder.com/job-details/senior-ai-ml-engineer--3e687eab-e14c-42ba-8c8c-0d6fa386c190

## About the Role

Artificial Intelligence (AI), Memory Hardware, Python Programming/Scripting Language, MCP - Microsoft Certified Professional, Engineering Change Management, Licensing, Tax Returns, Benchmarking, Access Control, Caching, * Professional Experience: 5+ years of software engineering experience in Python with strong expertise in high-throughput asynchronous services (`asyncio`, FastAPI).

 * LLM & Agent Systems: 2+ years of hands-on experience building production LLM applications, agentic workflows, function calling, tool use, and multi-agent interaction protocols (e.g., LangChain, LangGraph, LlamaIndex, or custom frameworks).
 * Protocols & Architectures: Solid understanding of Model Context Protocol (MCP), RESTful API design, Agent-to-Agent (A2A) orchestration, and Knowledge Graph/RAG patterns.
 * Evaluation Methodologies: Demonstrated experience establishing automated LLM evaluation datasets, synthetic test benches, and scoring metrics for intent detection and tool selection.
 * Cloud & Data Stack: Experience with AWS cloud infrastructure (ECS/EKS, Lambda, SQS/Kafka), vector databases, and enterprise data warehouses (Snowflake).
 * Software Craftsmanship: Strong commitment to unit/integration testing, CI/CD automation, clear API documentation, and agile team collaboration.

Preferred Experience

Prior experience working with enterprise SaaS, tax/compliance technology, or complex multi-step workflow automation.

Hands-on experience with streaming architectures, semantic caching, and LLM latency optimization.

Familiarity with IAM/RBAC entitlement models and enterprise data security standards (SOC 2, ISO).
Must Haves
Deep experience in Python and high-performance asynchronous API frameworks (asyncio, FastAPI)
Hands-on experience developing LLM agents, tool calling, and multi-agent orchestration
(LangChain, LangGraph, or custom agent architectures)
Experience implementing Model Context Protocol (MCP) or Agent-to-Agent (A2A) integration layers
Experience with LLM evaluation pipelines, synthetic dataset creation, and latency/token optimization

## Description

* Agent Orchestration & Backend Development: Design, implement, and optimize multi-agent orchestration workflows (A2A), prompt templating, dynamic intent routing, and conversation memory using Python, `asyncio`, and FastAPI.
 * MCP & Tool Integration: Standardize and integrate Model Context Protocol (MCP) servers and tools across Avalara products (AvaTax, Returns, Tax Research, ECM, Licensing).
 * Evaluations & Benchmarking: Build automated evaluation pipelines (LLM-as-a-judge, deterministic assertions, harness evals) to benchmark agent intent accuracy, routing precision (HSR/FSR), and execution latency.
 * Gating & Entitlement Routing: Implement dynamic cohort-based feature gating and IAM-DS entitlement evaluation to support controlled beta rollouts and multi-tenant access controls.
 * Performance & Reliability: Optimize token consumption, response streaming, context caching, rate-limiting, and error handling for high-concurrency agent interactions.
 * Production Readiness & Observability: Build telemetry pipelines, structured logging, and monitoring dashboards across Datadog, Snowflake, and OpenTelemetry to track model accuracy and system health.