U.S. citizenship required due to government contract support.
Hands-on experience deploying ML or LLM systems in a DoD IL5 environment (Azure Government preferred).
7+ years in ML or applied AI, with experience owning models serving real traffic.
Deep expertise in Azure OpenAI, RAG, vector search, and evaluation metrics.
Ability to work in restricted environments without standard installation tools.
Responsibilities
Serve Azure OpenAI and Azure AI Speech models, optimizing for quality, latency, and cost.
Port RAG and HLM pipelines to Azure Government services, ensuring compliance with boundary restrictions.
Build evaluations for model accuracy and report findings to the customer.
Document model versions, prompts, and data lineage for Authorization to Operate (ATO).
Run Voice Capture using Azure AI Speech within IL5 constraints.
Develop access control and audit logging in collaboration with full stack engineers.
Benefits
Equity participation in the company.
Comprehensive benefits package including health and wellness programs.
Opportunities for professional growth and development.
Collaborative work environment focused on innovation.
Full Job Description
San Francisco / Los Altos, in-person 5 days a week, occasional travel to customer sites • Reports to the CTO • $200K to $260K base + equity About the role
You will make the HLM run on Azure OpenAI inside a Department of Defense/War Impact Level 5 (IL5) boundary, and prove it performs as well there as it does commercially. What you'll do
Serve the models. Stand up Azure OpenAI and Azure AI Speech in Azure Government, and tune them for quality, latency, and cost.
Port the pipelines. Move our RAG and HLM pipelines onto Azure Government services, with nothing leaving the boundary.
Prove it works. Build evals for accuracy, grounding, and refusals, and report results to the customer.
Own provenance. Model versions, prompts, approved images, and data lineage, documented for the ATO.
Bring voice. Run Voice Capture on Azure AI Speech inside IL5.
Secure retrieval. Build access control and audit logging with our full stack engineer.
What we're looking for
U.S. citizenship is required. This role supports U.S. government contracts that require it.
Hands-on experience deploying ML or LLM systems in a DoD IL5 environment (Azure Government preferred).
7+ years in ML or applied AI, owning models that serve real traffic.
Deep experience with Azure OpenAI or similar hosted models, RAG, vector search (such as Azure AI Search), and evals.
Comfortable in disconnected or tightly restricted environments where you cannot pip install your way out of a problem.
Able to pass a U.S. government background investigation. An active clearance is a plus.
Bonus
Speech and audio ML (ASR, voice synthesis).
Fine-tuning models on Azure OpenAI or Azure Machine Learning.
Iron Bank / Platform One container pipelines.
Apply with
An ML system you deployed in IL5: the models, the serving stack, and how you evaluated it. Nothing classified or CUI. Papers or repos welcome. How we hire
We hire for a growth mindset: outcomes over output, experiments by default, and moving the metric that matters.
Location: in-person, 5 days a week, across our San Francisco and Los Altos offices.
The expected base compensation for this position is below. Our total compensation package for full-time employees includes equity and a comprehensive benefits package.
