Описание

This isn't a research role. You'll ship production AI systems that serve real clients and handle real legal workflows.

Design and build the repeatable agent framework that powers Lexsy's COO, Lawyer, CTO, and Growth agents
Multi-model architecture (Claude, GPT, Gemini) with intelligent fallback, context management, and cost optimization
Build and maintain integrations with 30+ external services (Gmail, Slack, Notion, Calendar, financial APIs, legal tools) via MCP and direct APIs
Human-in-the-loop approval gates, data isolation, audit logging, and compliance enforcement at the framework level
Monitoring, alerting, reliability engineering for AI systems that serve paying clients 24/7
AI-powered client-facing features — cap table management, virtual data rooms, Word document generation and redlining, workflow automation, intelligent routing, and investor reporting tools

5+ years of software engineering experience with production systems at scale
Strong experience with LLM/AI integration in production (not just prototypes)
Full-stack capability: you can build the API, the integration, and the UI
Experience with Next.js, TypeScript, and PostgreSQL (we use Supabase)
Understanding of LLM reliability patterns — when to use AI vs. deterministic logic, how to handle hallucinations, context window management
Startup comfort — you've worked in environments where you own problems end-to-end
Strong debugging skills under production pressure

Experience building multi-agent systems or AI agent frameworks
Familiarity with MCP (Model Context Protocol) or similar tool-use architectures
Experience with cloud platforms (Railway, Vercel, AWS, or GCP)
Background in regulated industries (legal, healthcare, fintech) where guardrails matter
Open source contributions or public technical writing
Experience with real-time systems (webhooks, streaming, event-driven architectures)
Familiarity with financial/legal data structures (cap tables, equity instruments, corporate records) and document processing (Word/DOCX generation, redlining, version tracking)

Senior AI Engineer