Senior Software Engineer - AI Innovation

Worth AI, a leader in AI onboarding and underwriting, is looking for a talented and experienced Senior Software Engineer - AI Innovation to join our team. At Worth AI, we are on a mission to revolutionize decision-making with the power of artificial intelligence helping fintechs, lenders, payment processors, and financial institutions onboard small businesses faster, smarter, and more confidently. We’re building the infrastructure that powers real-time KYB, KYC/IDV, underwriting, and continuous risk monitoring at enterprise scale, and the Worth Score™ our unified credit score derived from 1,200+ data points across 700M+ SMBs.

As a Senior Software Engineer - AI Innovation, you won’t be wiring up demos you’ll be designing and shipping production agent systems that make consequential decisions on regulated financial data. Worth’s platform consolidates onboarding and underwriting into a single AI-powered system, and our agents read, reason over, and act on the messy, high-stakes signals that come with that domain. You’ll own the end-to-end lifecycle: architecting agent graphs, building the retrieval and tool layers they rely on, instrumenting them with evals and observability, and getting them deployed against SOC 2 / GDPR / CCPA guardrails. You’ll partner closely with our Chief AI Officer, applied scientists, product, and platform teams to turn agentic patterns into customer outcomes.

Responsibilities

• Design and ship multi-step agentic systems (planner/executor, tool-using, multi-agent, human-in-the-loop) that automate KYB, underwriting, case review, and risk monitoring workflows.
• Architect agent graphs in LangGraph (or comparable frameworks CrewAI, AutoGen, Claude Agent SDK) with explicit state, durable execution, retries, and safe fallbacks.
• Build and harden the retrieval layer powering our agents chunking strategies, hybrid search, reranking, and grounded citation across SoS filings, IRS records, bank data, and Worth’s 700M+ SMB graph.
• Own the eval stack: golden sets, offline regression suites, LLM-as-judge, online A/B and shadow evals, and red-teaming for jailbreaks, prompt injection, and PII leakage.
• Wire agents into Worth’s production systems via well-typed tools, MCP servers, and existing services (decisioning engine, case management, crosswalking). Treat tool surface area as a product.
• Drive production MLOps for agents: deployment, versioning, traffic shaping, cost/latency budgets, observability (traces, token spend, tool call success), and on-call playbooks for agent incidents.
• Partner with security, compliance, and legal to keep agents inside Worth’s SOC 2, GDPR, CCPA, and fair-lending posture — building from day one, not bolted on.
• Translate ambiguous product bets (“what if the underwriter had an AI co-pilot for this?”) into concrete agent designs, prototypes, and shipped features.
• Mentor engineers across the org on agent patterns, prompt engineering hygiene, eval discipline, and the failure modes of LLM systems.
• Stay ahead of the frontier new models, frameworks, and patterns — and bring back what actually works in production.

Technology Stack

• Languages & Runtimes: Python, Node.js, TypeScript
• Agent / LLM frameworks: LangGraph, LangChain, Claude Agent SDK, MCP, OpenAI SDK
• Models: Anthropic Claude, OpenAI, open-weight (Llama, Mistral) where appropriate
• Retrieval & Data: PostgreSQL, pgvector / vector DBs, OpenSearch, Kafka, Redshift, Redis
• Infra & Orchestration: AWS, Kubernetes (EKS), ArgoCD, Terraform
• Evals & Observability: LangSmith / Langfuse / Braintrust-style tooling, DataDog, custom eval harnesses