Software Engineering Co-Op Internship

PetsApp; New York (NY); On-site; USD 70,000 - 90,000; Full time

Job summary

PetsApp in New York is looking for a role focused on prompt engineering and model evaluation in the financial domain. You'll work on testing and improving AI-driven document understanding to facilitate financial advising services.

Your responsibilities will include evaluating AI extraction performance, maintaining financial document datasets, and developing internal tools that enhance extraction quality. Ideal candidates have programming experience and a keen interest in structured data.

Qualifications

• Strong attention to detail and the ability to analyze model behavior.
• Comfortable working with structured data formats like JSON and CSV.
• Some programming experience is preferred.

Responsibilities

• Use tools like Reducto Studio to evaluate AI extraction performance.
• Build and maintain a curated test dataset of financial documents.
• Develop internal tools for automation and quality tracking.

Skills

Attention to detail

Programming experience

Experience with structured data

Experimenting with LLMs

Tools

Ruby

Bash

Adobe Acrobat

Job description

About LEA

LEA is building the document understanding API for wealth management. We help financial advisors and RIA (Registered Investment Advisor) firms unlock the power of their client documents using AI-powered data extraction and workflow automation.

By structuring data from PDFs like brokerage statements and financial plans, we enable use cases like proposal generation, onboarding, and client insights — integrating directly into systems like Orion, Morningstar, or internal advisor portals.

We’ve closed our seed round and are growing our team to refine our extraction engine and bring powerful, AI-driven workflows to the wealth management industry.

What You’ll Work On

This role is a mix of prompt engineering, model evaluation, internal tooling, and data curation. You’ll support our efforts to test and improve how we extract structured financial data from unstructured documents.

While most of the work will focus on testing and refining model outputs, you’ll also spend ~25% of your time coding — building internal tools to streamline this process.

Key Areas of Focus

Prompt Engineering & Model Evaluation (60%)

• Use tools like Reducto Studio to test and evaluate AI extraction performance
• Run documents through different prompts & schemas to compare output quality
• Identify inconsistencies and iterate on prompts for more accurate results
• Log test outcomes, manage prompt versions, and maintain evaluation docs

Dataset Management & Redaction (15%)

• Build and maintain a curated test dataset of financial documents
• Editing documents using tools like Adobe Acrobat
• Organize files and metadata into structured directories for version-controlled testing

Internal Tooling & Coding (25%)

• Build small internal tools (e.g. scripts, dashboards, validators) to:
  • Automate testing runs across prompts or schemas
  • Compare JSON outputs and flag anomalies
  • Track evaluation history across document versions
• Technologies might include: Ruby, Bash, simple web apps, JSON tooling, etc.

This part of the role is where you can flex your technical skills— and help us improve how we measure and enhance the extraction quality at scale.

You Might Be a Good Fit If You:

• Love experimenting with LLMs and prompt design
• Have strong attention to detail and enjoy analyzing model behavior
• Are comfortable working with structured data (JSON, CSV) and document formats (PDF)
• Have some programming experience and want to apply it to real-world AI systems
• Are organized, communicative, and excited about startup life

Bonus Points For:

• Experience writing scripts or simple internal tools (Ruby preferred)
• Familiarity with data extraction, NLP, or LLM tools (e.g. OpenAI, Claude, Reducto, etc.)
• Knowledge of document processing, evaluation frameworks, or prompt versioning
• Interest in fintech, wealth management, or infrastructure-layer AI