AI Agent Developer

Braintrust

Seattle, WA(remote)

JOB DETAILS
SKILLS
Application Programming Interface (API), Architectural Services, Artificial Intelligence (AI), Artificial Intelligence (AI) Agents, Automation, Biotech and Pharmaceutical, Competitive Analysis/Strategy, Consulting, Customer Relations, Database Design, Develop and Maintain Customers, Electrical Wiring, English Language, Gap Analysis, Health Science, Image Optimization, MCP - Microsoft Certified Professional, Mergers and Acquisitions, Network Connectivity, OAuth, Ontology, PostgreSQL, Project/Program Management, Prototyping, Python Programming/Scripting Language, REST (Representational State Transfer), Storyboards, Structured Data, System Architecture, User Interface/Experience (UI/UX)
LOCATION
Seattle, WA
POSTED
Today
Company
Braintrust is a global talent network that connects top independent professionals with leading companies for high-quality, flexible work. We help organizations hire skilled talent faster while giving professionals access to vetted opportunities with innovative teams.

Job description

AI Agent Developer


About industryGENOME

industryGENOME (iG) is a health sciences market intelligence platform providing bottom-up

industry analysis, financial and deal metrics, and highly precise market maps for the

pharmaceutical services industry. iG is building a multi-product AI platform with two distinct

product lines:


DealBeans — iG’s AI engine for investment banking workflows. DealBeans ingests, classifies,

extracts, and synthesizes M&A documents into investment-grade deliverables (CIM/offering

memos, gap analyses, storyboards, slide instructions). DealBeans maintains its own Client

Knowledge Graph — a structured data layer built from client-specific source documents through

an automated ingestion pipeline. DealBeans is live and in production use at Crosstree Capital,

iG’s affiliated M&A advisory firm.

Strategic Intelligence Agents — AI-powered tools within the iG platform for non-investment-

banking use cases, including competitive landscape analysis, market mapping, and total

addressable market sizing. These agents operate directly against iG’s proprietary database (not

the DealBeans Client Knowledge Graph) and serve strategic consulting clients such as PE

firms, pharmaceutical service companies, and sponsor/innovator pharmaceutical companies.

A private MCP server will connect iG’s proprietary data engine to both product lines, providing

agents with a secure, standardized interface for accessing proprietary datasets and internal

business tools.


The Role

You will be the lead engineer responsible for productizing a working AI agent prototype into

commercial-grade software. The CIM Engine prototype — which ingests, classifies, extracts,

and synthesizes M&A documents into structured deliverables — is complete and validated. Your

job is to harden it, extend it, build the client-facing layer, and scale it beyond a single user.

You will be onboarded into DealBeans by the founder/architect, who designed the agent

workflows and prompt architecture. Once the product handover is complete, you will report

directly to Frank, Ontology Manager and de facto President of industryGENOME, consulting the

founder for clarifying questions and architectural direction on DealBeans-specific work. This is a

high-autonomy role with significant ownership of the technical stack.


What You’ll Build

First 90 Days

• Production hardening of the existing N8N + Claude API pipeline (async processing, error

handling, retry logic, monitoring)

• Web UI for project management, artifact review, and workflow triggering (Supabase

backend already stores all necessary data)

• Multi-client architecture — the prototype serves one client; you’ll extend it to serve many

• Image generation pipeline optimization (Gemini API integration via OpenRouter)

• Content quality improvements to the prompt-driven generation chain

Months 4–12

• Private MCP server exposing iG’s proprietary data to AI agents across both product lines

• Strategic intelligence agents within iG (competitive landscape, market mapping, TAM

sizing) operating against iG’s proprietary database

• Additional DealBeans engines consuming the Client Knowledge Graph (Pitch Engine,

Market Landscape, Competitive Assessment)

• PPTX assembly automation and template management

• API layer for external client access


REQUIRED QUALIFICATIONS

• 3+ years building and deploying web applications or data pipelines in a production

environment — not just prototypes or demos

• Strong Python and JavaScript/TypeScript proficiency — you'll work across both languages

daily

• Database design and management — PostgreSQL required; Supabase experience a plus

• API development and integration (REST, webhooks, OAuth) you'll connect multiple third-

party services and build internal endpoints from scratch

• Frontend development skills (React/Next.js) building a web UI is part of the role

• Experience wiring together multi-step automated workflows — whether through tools like

N8N, custom Python pipelines, or equivalent; the pattern matters more than the specific

tool

• Working experience with LLM APIs (OpenAI, Anthropic Claude, or Google Gemini) you

don't need to be an AI researcher, but you should have called these APIs, handled

structured outputs, and shipped something that uses them

• Fluent English — you'll work directly with the founder on product decisions and domain-

specific content


NICE TO HAVE

– RAG pipeline experience (vector search, embeddings, retrieval)

– MCP (Model Context Protocol) server experience

– PDF extraction or document parsing pipelines

– Financial services or document intelligence background

– N8N specifically


OUR STACK

You'll be working with these technologies from day one. You don't need prior experience with

every item, but you should be comfortable picking up unfamiliar tools quickly.



Orchestration- N8N — primary workflow engine; most logic lives here

AI- Claude API (Sonnet 4.6, Opus 4.6) via OpenRouter

Image Generation - Gemini via OpenRouter

Database - Supabase (PostgreSQL + file storage)

File Storage - Google Drive API

Agent Protocol - MCP (Model Context Protocol)

Languages - JavaScript (N8N code nodes), Python (utilities and scripts)

Frontend - React or Next.js + Supabase Auth


Engagement Details

Location: Remote — global. Must be available during overlapping hours with US Eastern time

(minimum 4 hours daily overlap). Target regions: Latin America, Eastern Europe.



About the Company

B

Braintrust