Applied AI Engineer

BRATHON; New York (NY); On-site; USD 180,000 - 230,000; Full time

Job summary

BRATHON is seeking an Applied AI SME to design and scale GenAI workflows within a production-grade AI platform for Lending. The role focuses on building enterprise-grade AI-based document ingestion and data extraction capabilities with traceability, confidence scoring, and human-in-the-loop review.

The ideal candidate has a strong math background, hands-on experience with live systems, and expertise in agentic workflows, LLM orchestration, and regulated environments.

Qualifications

• Must have hands-on experience building and deploying GenAI-powered solutions.

Responsibilities

• Design and evolve GenAI workflows across Lending business lines.
• Develop an enterprise-grade AI-based document ingestion and data extraction capability with traceability and human-in-the-loop review.
• Build AI-powered assistants embedded in Lending systems using agentic workflows.
• Deliver automated content and deck generation workflows for reporting and approvals.
• Provide expert advice on GenAI architecture including model selection and orchestration patterns.

Skills

Math background

Applied AI SME

Production mindset

Regulatory mindset

Audit awareness

Education

Advanced degree in CS/Math/Stats

Tools

PyTorch

Langchain

RAG

LLMOps

Document ingestion

Job description

• Strong background within Machine Learning
• Everything withing AI starts here
• Strong within Python (NOT looking for a software developer who has learned how to use models. Instead, looking for a core math background, machine learning, data model extraction and/or trading models.
• Pytorch
• Langchain
• Core AI Tools
• Document Extraction Experience and comfortable enhancing the model to produce better extraction
• RAG experience is a MUST
• Agentic Workflows Exp
• Ability to orchestrate workflows of a system. Someone who has applied their knowledge into apply agentc workflows - move through steps from beginning to end. Do transformation on it and produce a result
• Llm OPS
• NOT LOOKING FOR A SOFTWARE DEV WHO learned how to use models
• Building production systems and not just trying to learn something
• Someone who has applied knowledge into how EFFECTIVELY they can apply agentic workflows
• Must have professional experience working with a LIVE system, opposed to someone doing this recreationally
• Financial Services is a HUGE plus

Plus:

• Engineering Exp front to back
• Build workflows, examine systems, suggest solutions to it using data flows and gen AI, etc

Fixed Income or Institutional Lending domain experience.

• Experience working in regulated environments with strong audit and control requirements.
• Familiarity with enterprise security, data governance, and entitlement models.
• Frontend experience is beneficial (Angular or React)

D2D;

• Design and evolve reusable GenAI workflows used across Lending business lines.
• Develop an enterprise grade AI-based document ingestion and data extraction capability, including traceability, confidence scoring, and human-in-the-loop review.
• Build AI-powered assistants embedded in Lending systems using agentic workflows.
• Deliver automated content and deck generation workflows for reporting and approvals.
• Provide expert advice on GenAI architecture including model selection, orchestration patterns, and evaluation strategy.
• Establish LLMOps practices: extraction accuracy, assistant reliability, prompts management, and audit monitoring.
• Design and implement controls for entitlements, PII handling within open-source models in a regulated environment.
• In the role you are expected to act as a hands-on technical expert, and it has a clear path to becoming a platform owner responsible for shared GenAI standards across Lending.

Misc:

• They do not have this type of candidate on team today - they need this candidate to come in and be a SME on the team and be confident enough to bring their own two sense into the role.
• AI Platform is being built from scratch
• Need this candidate to bring knowledge of how to chain workflows and how to enhance extractions
• Stakeholders
• Strats (heavy math background) + Operations Users + Business Users
• Docs can go from 1 page to 300+ pages (PDFs)

Redflags:

• Developer who simply is using AI

Ideal Candidate:

• Applied AI SME - how can come in and say " THIs is how it needs to be done, this is what I will need to do it, this is my timeline, etc."
• Practitioner is someone who not just someone giving out straight answer.