Artificial Intelligence Engineer

Stelvio Inc.; Austin (TX); On-site; USD 140,000 - 190,000; Full time

Job summary

Stelvio Inc., Austin-based technology company, is seeking an experienced AI/ML Engineer to own end-to-end model lifecycle from experimentation to production.

The ideal candidate will have 5+ years in ML or related software development, deep experience with PyTorch/TensorFlow, and a track record of deploying models in production. This hands-on role emphasizes model engineering over prompts or API integration.

Qualifications

• Experience with training, fine-tuning or modifying deep-learning models.
• Experience releasing a model from research to production.
• Ability to build evaluation methods and judge model quality.
• Experience deploying model-driven systems in customer-facing environments.
• Strong Python with PyTorch, JAX or TensorFlow.
• Experience improving GPU utilization, inference speed or training efficiency.
• Familiarity with CUDA ecosystem and high-performance workloads.
• Ability to balance accuracy, latency, throughput, reliability and infra cost.
• Ability to operate independently where technical path is not predefined.

Responsibilities

• Own ML model lifecycle from experimentation through production release.
• Train, fine-tune, or modify deep-learning models for real-world use.
• Release models into production and monitor performance.
• Develop evaluation methods to inform quality decisions.
• Deploy model-driven systems in customer-facing environments.
• Optimize GPU usage and inference efficiency.
• Work with PyTorch, JAX, or TensorFlow.
• Balance trade-offs among accuracy, latency, throughput, and cost.
• Operate independently in undefined technical paths.

Skills

Training/fine-tuning deep-learning

Model lifecycle ownership

Model release to production

Model evaluation methods

Deploying model-driven systems

Python with PyTorch/JAX/TensorFlow

GPU utilization optimization

CUDA ecosystem familiarity

trade-offs: accuracy/latency/cost

Independent execution in undefined路径

Job description

I’m supporting an Austin-based technology company with the appointment of a highly experienced AI/ML Engineer.

This is a hands-on model-engineering position - not a prompt-engineering role or an opportunity focused primarily on connecting applications to third-party AI APIs.

The company needs someone who has taken meaningful ownership of machine-learning models across several stages of their lifecycle: experimentation, training, adaptation, evaluation, performance improvement, release and production operation.

The expected level is five or more years in machine-learning, research engineering or technically comparable software development. More important than the number of years, however, is evidence of what you have personally built.

Your background should demonstrate several of the following:

• Training, fine-tuning or otherwise materially modifying deep-learning models
• Releasing a model or taking one from research prototype into real-world use
• Building evaluation methods and making informed decisions around model quality
• Deploying model-driven systems within a commercial, customer-facing environment
• Strong Python experience with frameworks such as PyTorch, JAX or TensorFlow
• Practical experience improving GPU utilization, inference speed or training efficiency
• Familiarity with the CUDA ecosystem and the constraints of high-performance workloads
• Making trade-offs between accuracy, latency, throughput, reliability and infrastructure cost
• Operating independently in an environment where the technical path is not already defined

Particular interest will be given to engineers who can point to something tangible they have created: a model, repository, technical paper, product capability, optimization project or system operating in production.

Relevant specialist experience could include transformer architectures, language models, multimodal systems, distributed training, model compression, quantization, compilation, optimized serving or lower-level GPU performance work.

Further information about the company, working arrangement and technical objectives will be shared selectively during an initial confidential conversation.