Machine Learning Engineer, Applied AI - Deployed

Role Overview

Machine Learning Engineer, Applied AI - Deployed at Brain Co., a company building AI-native operating systems for large, regulated institutions.

Company Overview

Brain Co. builds AI-native operating systems for large, regulated institutions. Each system is built for a specific industry, powered by agents that push real workflows forward. The company is powered by Atlas, a proprietary platform that keeps customers in control, secure by design, and never locked into one model. Brain Co. is entering its next phase of production deployments on a national scale with an elite team built from Palantir, Google, Meta, and Nvidia, and a growing footprint across government, insurance, health, and financial services.

Role Purpose

You will transform ambiguous institutional problems into production ML systems that automate critical workflows in regulated industries. Your work bridges the gap between impressive model demonstrations and systems that institutions stake their processes on, handling the full lifecycle from problem definition through production deployment where real decisions get made.

Key Responsibilities

System Architecture & Composition

• Own AI systems end-to-end, from training through production behavior.
• Build composite systems that chain vision transformers, segmentation models, VLM reasoning, and rule engines into integrated solutions.
• Determine which components—fine-tuned segmentation models, VLMs, or rule engines—are optimal for each problem and compose them into systems more accurate than any single model.
• Engineer for production reality, navigating accuracy, latency, cost, and reliability in environments far messier than any benchmark.

Problem Definition & Data

• Turn ambiguity into shipped systems, moving from no problem statement, no labeled data, and no agreed definition of success to well-posed ML problems and production deployments.
• Work directly with the institutions you serve—permit reviewers, underwriters, compliance officers—to understand how decisions actually get made and ensure your systems change how the work gets done.
• Design data pipelines and labeling strategies for institutional-grade accuracy on complex, multimodal documents.

Document Understanding & Modeling

• Build models that handle dense, multimodal documents—blueprints, site plans, policy stacks, contracts, clinical records—that break off-the-shelf models.
• Build custom vision model pipelines that check blueprints against building codes at 95%+ accuracy.
• Develop agents that untangle policy stacks to reveal coverage gaps.
• Create systems that predict from clinical records whether a patient is on their care path.

Evaluation & Quality Assurance

• Build evaluation suites and failure-mode taxonomies rigorous enough to earn institutional sign-off.
• Treat evaluation as a product discipline where regulators must trust your system.
• Own the problem of credit assignment when composite pipelines fail—determining which component is responsible.
• Leverage verified, ground-truth outcomes from production deployments as reward signals for model improvement.

Agent Development & Learning

• Work at the research frontier with production stakes, applying LLMs, RL fine-tuning, and agentic systems where output is a decision an institution acts on.
• Design data, evals, and training loops to build and fine-tune agents on real-world verified outcomes.
• Help agents learn from real work and ground-truth feedback generated by production deployments.

Institutional Intelligence & Feedback Loops

• Work on systems where every verified correction improves the system twice: the corrected fact percolates to every application, and the system that builds the intelligence learns to build it better.
• Own both the correction loop and the learning loop for institutional intelligence that compounds.

Knowledge Sharing & Standards

• Raise the bar across the company through design reviews, the internal paper club, and the shared playbook for AI systems institutions can trust.

Qualifications & Experience

• Deep understanding of how machine learning actually works—the philosophy underneath, not just the tooling.
• Hard-won instincts for squeezing the most out of LLMs and agentic systems, including prompting, fine-tuning, tool use, and reasoning.
• Experience knowing when a fine-tuned segmentation model beats a VLM, when a rule engine beats both, and how to compose all three.
• Ability to understand what a loss function really optimizes, how generalization breaks under distribution shift, and why evaluation is where systems quietly go wrong.

Skills & Competencies

• Expertise in LLMs and agentic systems.
• Proficiency with vision transformers, segmentation models, and VLMs.
• Skill in RL fine-tuning and model composition.
• Ability to treat frontier models as components to be measured, pushed, and engineered rather than as magic.
• Comfort with underspecified problems where the problem, data, and definition of success all have to be invented at once.
• Capability to apply research-frontier techniques in production environments with institutional-grade accuracy requirements.

Additional Information

Every project here ships to production and is expected to create measurable customer value and impact. You will work alongside exceptional peers on some of the hardest problems in applied AI.