Ai Product Engineer (Agent & Integrations)

Ensemble Ip; Barcelona; Hybrid; EUR 65,000 - 100,000; Full time

Benefits offered by this job

Remote work option

Co-working in Gracia

Job summary

Ensemble IP LLC is seeking an AI Product Engineer in Barcelona to expand Duet's agent harness. You will design and implement MCP access, integrate new data sources, and author prompts and skills for an evolving AI platform.

The role emphasizes incremental delivery, data source integration, and collaboration with a small, driven team. You will work on context management, production agent loops, and tooling integrations, with a focus on shipping value in each sprint.

Qualifications

• 3+ years of professional software engineering experience with strong TypeScript/Node backend skills.
• 12+ months building LLM agents and integrating multiple tools and data sources.
• Demonstrated ability to decompose work into incremental, value-delivering milestones.
• Excellent written and verbal communication and collaboration in a small team.

Responsibilities

• Extend Duet's agent harness by integrating new data sources and MCP access.
• Write thin TypeScript connectors and wire data sources into the agent as tools.
• Prototype, iterate prompts and tool definitions to improve agent performance.

Skills

TypeScript/Node

LLM agents

English fluency

incremental delivery

prompt engineering

Tools

MCP

Redis

APIs integration

Git version control

Job description

About Ensemble IP

Ensemble IP is a U.S.-based patent search and software company founded by veterans of the professional patent search industry. Our team of experienced patent analysts delivers patentability, freedom-to-operate, invalidity, and landscape searches to law firms, in-house IP teams, R&D organizations, and inventors who rely on accurate prior art to make high-stakes decisions.

We pair that domain expertise with our own software. Quartet, our AI-enabled patent search platform, integrates AI and Boolean search techniques to help professionals find better prior art, faster. Duet extends that platform into agentic AI. Because our analysts use these tools every day, engineers work side by side with the people who know the problem best.

We have a small development team in Barcelona and offer both remote and co-working options. Our co-work location is in Gracia.

About Our Products

Duet is an agentic AI application for anyone needing to find, understand, or interact with patent data. Users ask natural-language questions about inventions, prior art, portfolios, and claims. Duet searches a proprietary patent corpus, reads documents, and returns grounded, actionable answers.

Duet sits in the Ensemble IP ecosystem alongside our traditional patent search platform Quartet. Duet implements a cutting-edge agent harness to use the resources made available by Quartet via tools, skills, and expert subagents.

This is applied AI engineering in production, not model training or demos.

The role

We're looking for an AI Product Engineer who efficiently ships outcomes. You'll receive a user problem, a rough capability, or a direction like “expose this through MCP”, not necessarily accompanied with a full interface design. Your job is to chart a path to the end-state, sequence the unknowns so you're never blocked on a single big bet, and put working functionality in the users’ hands at every stage. We innovate and move fast. The focus is on thin vertical slices that compound to produce impressive features.

Concretely, that means extending Duet's agent harness: integrating new data sources, building MCP access to Duet functionality, and authoring curated prompts and skills. The AI Product Engineer needs to bring their expertise to these processes to implement robust and cutting-edge solutions with incremental wins along the way.

How We Expect You To Work

Start from outcomes, not design requirements. We are obsessed with delivering value to our customers. You translate user needs into solutions with a staged delivery plan that brings value to the user in each sprint.

Be curious and engaged. We are a small company that values the voice of every team member. We expect every engineer to become familiar with the world of intellectual property and to bring your ideas and talents to the table so that we can create the best possible solutions for our customers.

Collaborate, challenge, and support your team members. We live the spirit of agile – we are radically transparent, we take introspection seriously, and we constantly adapt and improve. You will be joining a highly collaborative team that thrives on exchanging ideas, challenging each other, and learning from each other.

What You'll Work On

MCP server access

Design and implement our approach to expose Duet's capabilities via MCP.

Rapid data integration

Add new data sources by writing thin TypeScript connectors and wiring them into the agent as composable tools. You'll often get “we need source X in the agent” before every field and failure mode is defined. Take the lead to understand the data source and find the most efficient way to incorporate it into Duet.

Curated prompts & skills

Author and iterate on system prompts, tool designs, skill definitions, and the architecture that links these resources together.

Agent platform work

Extend and improve our production agent loop. You’ll work on context management, streaming, Redis locks, billing, presentation to the user, etc.

What We’re Looking For

• Fluency in written and verbal English
• 3+ years professional software engineering experience, with strong TypeScript/Node backend skills
• 12+ months building LLM agents, with an emphasis on integrating orchestrating complex tools and data sources
• A track record of delivering outcomes efficiently
• Incremental delivery discipline - you decompose work so that each PR adds user value or represents a deliberate milestone that leads to the next piece of functionality
• Demonstrated prompt engineering as product work - structured prompts, iteration from failure modes, version control
• Experience integrating third-party APIs into an agent framework
• Excellent written and verbal communication with a focus on teamwork and collaboration
• mphasis on integrating orchestrating complex tools and data sources
• A track record of delivering outcomes efficientlyIncremental delivery discipline - you decompose work so that each PR adds user value or represents a deliberate milestone that leads to the next piece of functionality
• Demonstrated prompt engineering as product work - structured prompts, iteration from failure modes, version control
• Experience integrating third-party APIs into an agent framework
• Excellent written and verbal communication with a focus on teamwork and collaboration