SENIOR AI ENGINEER (ONSITE ROLE) at RADIANS INC

Memphis, Tennessee, United States -

Full Time

Start Date

Immediate

Expiry Date

04 Aug, 26

Salary

0.0

Posted On

06 May, 26

Experience

5 year(s) or above

Remote Job

Yes

Telecommute

Yes

Sponsor Visa

Skills

C#, .NET 8, Python, SQL, LLM Integration, Prompt Design, Data Pipelines, Microsoft Graph API, Blazor Server, Reliability Engineering, Document Extraction, API Integration, Azure, AWS, Prompt Engineering, SQL Server

Industry

Wholesale

Description

Description Role summary You are the primary builder for the AI Integration team's roadmap. You take scoped initiatives from the AI Adoption Analyst and Department Head, architect them, and ship them into production — end-to-end. That includes application code, the data pipelines that feed them, the integrations with our ERP and adjacent systems, and the observability to know they're working. This role is explicitly a hybrid: you'll write production C#/.NET application code the majority of the time, but you'll also write Python for data work, SQL for ERP integrations, and occasional infrastructure scripts. You're comfortable switching contexts and don't treat "that's not my stack" as a reason to stop. What you'll own · Application development — .NET 8 / C# against our existing platform (Arc on Blazor Server). Extend existing modules, build new ones, integrate with Prophet21 via both its API and direct SQL. · LLM integration work — production-grade Anthropic (and possibly OpenAI) API usage. Prompt design, tool use, conversation state, context management, cost tracking, fallback/retry behavior. Most of the team's tools have an AI reasoning layer somewhere. · Data pipelines — ingestion from email (Microsoft Graph), PDFs (PyMuPDF, vision models), and external vendors. Transformation and matching logic. Storage in SQL Server. · Quality and drift monitoring — as the person closest to the model calls, you own the metrics that show whether the AI is still producing good output. Accuracy dashboards, eval frameworks, regression checks when prompts or models change. · Reliability engineering — Windows services, background workers, error handling, logging discipline, on-call awareness. Not DevOps-heavy, but you don't ship and wash your hands. · Code review partner — collaborate on design with the Department Head and (as we grow) mentor the next engineer on the team. Requirements Must-have qualifications · 5+ years of production software engineering experience shipping features people use every day. · Strong proficiency in one of: .NET/C#, Python, Java/Kotlin, TypeScript/Node — with a willingness to cross-train into the others. Our stack is primarily .NET 8 and Python; prior .NET experience is a strong plus but a great engineer in another stack can ramp. · SQL proficiency: you can write non-trivial queries, read execution plans, and join across multiple sources without holding your nose. · At least one production system you've shipped that uses an LLM API (Anthropic, OpenAI, Cohere, Mistral — doesn't matter which). You understand the failure modes: rate limits, hallucinations, cost escalation, context window management, tool use loops. · Comfortable working with data that isn't clean: document extraction, fuzzy matching, multi-source deduplication, schema mismatches. · On-site at Radians Memphis HQ, full-time. ·Nice-to-have qualifications · Microsoft Graph API experience (Teams integrations, Outlook integrations, SharePoint). · Prophet21 or similar distribution/manufacturing ERP integration experience. · Prompt engineering or eval-framework experience (Langsmith, Promptfoo, internal harnesses). · Experience with document intelligence pipelines (intake ? extraction ? classification ? downstream action). · Azure or AWS cloud experience. · Open-source contributions or public technical writing.

Responsibilities

The role involves architecting and shipping end-to-end AI integrations, including application code, data pipelines, and ERP integrations. Responsibilities include managing LLM API usage, monitoring model drift, and ensuring system reliability.