Artificial Intelligence Engineer

About TwinCounsel

TwinCounsel is an early-stage legal tech startup building an AI delegation platform for solo and small law firms. Founded by YC S17 alumni (Indivio, acquired by Cloudinary), we're a proven team applying our experience in automation and AI to transform how lawyers work.

Our platform connects to a lawyer's email inbox, automatically builds knowledge about their cases, and lets them delegate complex work in plain English like having an expert paralegal who never forgets anything. We've already validated 90% time savings with our design partner.

With founders based in the US and a growing engineering team in Taiwan, we're building world-class technology that transforms how lawyers work.

Our Engineering Philosophy

We believe great engineers care deeply about how they build and why they build. Our culture values both the craft of engineering excellence and the care for the users and problems we're solving. That's the TwinCounsel way.

The Role: Orchestrating Intelligence

As one of our first engineers, you'll shape both our product and engineering culture. You'll be the conductor designing how our AI agent thinks, reasons, and makes decisions. You'll own the entire intelligence layer: from understanding what a lawyer is asking for, to retrieving the right context from their case history, to ensuring every generated document is accurate and auditable.

This is a hands-on role for someone who thinks like a scientist. You'll form hypotheses about why the model behaves a certain way, design experiments to test them, and iterate until the AI reliably helps lawyers do their best work. You'll measure success not by shipping features, but by improving accuracy, trust, and user outcomes.

Key Responsibilities

• Design and refine agent workflows how the AI understands requests, retrieves context, and generates responses
• Build and optimize RAG pipelines for accurate, context-aware retrieval from case documents
• Develop evaluation frameworks to measure and improve AI accuracy, citation quality, and task completion
• Engineer prompts and fine-tune model behavior for legal domain tasks
• Design intent classification systems to route lawyer requests to the right workflows
• Build auditability features ensuring every AI-generated fact traces back to its source
• Collaborate with Automation Engineers to define clear interfaces between AI reasoning and tool capabilities
• Analyze production errors and continuously improve agent reliability

What We're Looking ForTechnical Foundation

• 2+ years of experience in ML engineering, NLP, or AI application development using Python or similar object-oriented languages
• Strong Python skills and comfort with ML/data science tooling
• Experience with LLM APIs (OpenAI, Anthropic, etc.) and prompt engineering
• Familiarity with RAG architectures, vector databases (Pinecone, Weaviate), and embedding strategies
• Experience building evaluation frameworks and measuring model quality
• Understanding of SQL databases and basic backend development

Nice to Have

• Experience with agent frameworks (LangChain, LlamaIndex, AutoGen, CrewAI)
• Background in NLP research or language model fine-tuning
• Experience with workflow orchestration tools (Temporal)
• Familiarity with legal tech or professional services automation

Human Qualities

• Craftsmanship: Pride in the details your code, your tools, and the systems you build
• Empathy for the Problem: Curiosity to understand how your work improves the experience of others, even when you don't meet them directly
• Learning Mindset: Eager to master both the technical how and the business why
• Test Obsession: Belief that untested code is unfinished code quality is non-negotiable
• Team Player: Collaborative, thoughtful, and motivated to contribute to a positive engineering culture

You'll Love This Role If...

• You're fascinated by how LLMs reason and want to make them better at real tasks
• You get excited about designing evaluation frameworks and measuring AI quality
• You think like a scientist forming hypotheses, running experiments, iterating based on data
• You enjoy the puzzle of why did the model do that and systematically fixing it
• You want to shape how AI agents work, not just use them

What Makes You Stand Out

• You can explain complex AI behavior in plain language to non-technical teammates
• You've debugged model behavior in production and improved real-world accuracy
• You have opinions about evaluation methodology and can defend them
• You're curious about the legal domain and excited to become a domain expert
• You already use AI tools (Claude, ChatGPT, Gemini, Grok, Copilot) to accelerate your own research, evaluation design, and iteration workflows
• Experience working on remote or cross-cultural teams
• Background in B2B SaaS development

Working Across Timezones

Our US-based founders are 16 hours away from our Taiwan engineering team. This means async communication is essential not optional. You'll need strong written English skills to document decisions, explain technical approaches, and collaborate effectively without real-time meetings.

We value engineers who over-communicate in writing, think through edge cases before asking questions, and can work autonomously while staying aligned with the team.

AI-Native Engineering

We're building AI that lawyers trust to do real work. We want engineers who already trust AI to do real work alongside them.

Our team uses AI tools daily for coding, debugging, documentation, research, and problem-solving. We expect you to have your own workflow with tools like Claude, ChatGPT, Gemini, Grok, Copilot, or Cursor, and to have developed opinions about when AI helps and when it gets in the way. This isn't about replacing engineering judgment; it's about amplifying it.

Compensation

• USD $48,000$72,000 / NTD 1,500,0002,200,000 + equity