The Role
Own the AI portion of a logistics customer portal, taking an existing assistant from prototype to production. Responsibilities include building secure RAG and tool-calling systems, enforcing multi-tenant data isolation, developing evaluation pipelines and guardrails, operating Azure AI infrastructure, integrating predictive models, and monitoring quality, safety, usage, and cost. The role also develops AI features such as briefings, support assistance, search, and recommendations while collaborating with data scientists, client architects, and security teams.
Summary Generated by Built In
Established in 1997, Softeq was built from the ground up to specialize in new product development and R&D, tackling the most difficult problems in the tech sphere. Now we've expanded to offer early-stage innovation and ideation plus digital transformation business consulting. Our superpower is to deliver all of this under one roof on a global scale. So let's get started and build a better future together!
We're looking for a Senior AI Engineer to join our team and contribute to its continued growth.
We are building a new customer portal for a large logistics company. Business users at the company's customers will use it to track shipments, see delays and ETAs, manage invoices and documents, and get reports.
An AI assistant answers customers' questions about their own operational and financial data. A working prototype exists; our job is to take it to production for external customers.
You will own the AI part of the product: make the assistant safe, accurate and measurable, and build the next AI features on top of it.
An AI assistant answers customers' questions about their own operational and financial data. A working prototype exists; our job is to take it to production for external customers.
You will own the AI part of the product: make the assistant safe, accurate and measurable, and build the next AI features on top of it.
What you will do
- Take the existing AI assistant from prototype to production: retrieval with Azure AI Search, tool calls over entitlement-scoped APIs, prompt and model versioning.
- Design narrow, typed tools for the assistant so that every answer respects the customer's data entitlements. The customer is identified from the authenticated session, never from the model's output.
- Ground answers in certified metrics from the lakehouse semantic layer, and show users where an answer comes from.
- Build evaluation pipelines: golden sets, groundedness and correctness metrics, tool-call accuracy, regression runs in CI that block a bad release.
- Run the AI gateway on Azure API Management: quotas, logging, content safety, model routing and cost control.
- Implement guardrails: prompt-injection defense, PII redaction, safe handling of retrieved content, human confirmation for sensitive actions.
- Build new AI features: daily briefings, AI-assisted support and knowledge search, personalized recommendations.
- Work with the Data Scientist to put predictive models (delays, congestion, cost) into the product: model serving, monitoring and drift alerts.
- Monitor the assistant in production: tracing, user feedback, quality and cost dashboards.
What we expect
- 3+ years building LLM applications in production for external users: RAG, function / tool calling, evaluation. Internal tools or pet projects alone are not enough.
- Strong Python in production: FastAPI or similar, async, tests.
- Hands-on experience with evaluation of LLM systems: you can show how you built a golden set, which metrics you tracked, and how results gated releases.
- Practical AI security: prompt injection (direct and indirect), data leakage between customers, least-privilege tools.
- Experience with multi-tenant data isolation: you know why a customer filter must live in the server or the data layer, not in the prompt.
- Good SQL and understanding of business metrics: you can write the query behind "on-time delivery % per customer last month" and know where the metric definition should live.
- Clear written and spoken English; comfortable explaining design trade-offs to client architects and security teams.
Nice to have
- Azure OpenAI, Azure AI Search, Azure API Management (AI gateway).
- Databricks: Mosaic AI, Model Serving, MLflow, Unity Catalog, metric views.
- MCP servers in production, with authentication and per-tool authorization.
- Agent frameworks (LangGraph, Semantic Kernel or similar) and an opinion on when not to use them.
- Observability for LLM apps (Langfuse, OpenTelemetry, App Insights).
- Logistics, transportation or supply-chain domain.
Our tech stack
Azure OpenAI · Azure AI Search · Azure API Management · Python / FastAPI · Azure Databricks (Delta, Unity Catalog, Lakeflow) · PostgreSQL / Lakebase · React / TypeScript · GitHub Actions
Softeq communicates only from @softeq.com email addresses. We never request payments or fees for any reason during hiring — including trainings or courses to be completed, equipment, onboarding, or background checks — and we will not ask for banking information, cryptocurrency or gift cards. If you receive a message from any other domain or requesting payment, do not respond and report it to [email protected]
Skills Required
- 3+ years building production LLM applications for external users, including RAG, function or tool calling, and evaluation
- Strong production Python experience, including FastAPI or similar frameworks, asynchronous programming, and testing
- Hands-on experience evaluating LLM systems, including golden sets, tracked metrics, and release gates
- Practical AI security experience, including prompt-injection defense, data-leakage prevention, and least-privilege tools
- Experience implementing multi-tenant data isolation and server- or data-layer customer filtering
- Good SQL skills and understanding of business metrics
- Clear written and spoken English; ability to explain design trade-offs to client architects and security teams
- Experience with Azure OpenAI, Azure AI Search, and Azure API Management
- Experience with Databricks Mosaic AI, Model Serving, MLflow, Unity Catalog, or metric views
- Production experience with MCP servers, authentication, and per-tool authorization
- Experience with agent frameworks such as LangGraph or Semantic Kernel
- Experience with LLM application observability tools such as Langfuse, OpenTelemetry, or App Insights
- Logistics, transportation, or supply-chain domain experience
Am I A Good Fit?
Get Personalized Job Insights.
Our AI-powered fit analysis compares your resume with a job listing so you know if your skills & experience align.
Success! Refresh the page to see how your skills align with this role.
The Company
What We Do
Softeq is a custom hardware & software development company. We provide full-stack development services under a single roof: embedded software, mobile, web, and desktop apps; devices, IoT and Cloud solutions. Headquartered in Houston, TX, we work with regional startups and global corporations.
.jpeg)
