AI Developer

EXANTE • Georgia, GA

2d • Hybrid

About The Position

We are a global wealth-tech company powering the next generation of trading. We deliver cutting-edge centralised trading solutions and robust B2B financial infrastructure. Our proprietary trading platform offers clients seamless access to stocks, ETFs, bonds, futures, options, swaps, funds, and currency pairs- all within a single, multi-currency account.
We believe in the boundless flow of capital and information. Whilst developing EXANTE, we saw technology as a means to improve global connectivity of the financial services industry. Over the years, we have earned a reputation for providing superior service and for getting the results our clients and customers want.
Most importantly, people have come to know us for our technological insights and focus on privacy. Our inspiration stems from the talent and ambition of our clients. We are offering an ecosystem of financial products and services that would provide the most tailored assistance to traders.
Everything at EXANTE starts with our people. We bring together 600+ minds from 65 nationalities across 70 locations. We place trust in our team with real autonomy - the freedom to drive change across products, processes and everything in between. That trust fuels innovation, and the demands of the markets build resilience; we adapt, push forward, and we look after each other while we do it.
That freedom is matched by real investment in growth: we back our people with ongoing learning, hands-on development and the kind of stretch opportunities that turn good careers into great ones. Here, progression isn’t something you wait for - it’s something you’re actively supported to pursue.
We are building the AI layer of our sales organisation: an agentic system used daily by ~190 relationship managers across 14 regions, working against our CRM, Jira and internal knowledge base. It answers questions on clients and portfolios, delivers briefs, manages tasks and suggests the next best action.
Next up: real coaching against a manager's targets and sales methodology, AI role play, proactive task suggestion driven by data signals, deeper integrations, and the evaluation infrastructure that tells us any of it works.
Small team, high autonomy, no groomed board. You take a scope, build it end to end and own how it behaves in production.
Stack: TypeScript / Node 22, Express 5, Drizzle ORM + PostgreSQL, Redis, Slack Bolt, Vertex AI (Gemini), Langfuse, Vitest, GitLab CI, Docker. React 19 + Vite + Tailwind/shadcn on the frontend. Python for evals, data work and glue. Roughly 80/20 backend to frontend — and you are not expected to know all of it, you are expected to get up to speed on your own.

Requirements

• Agent engineering — you have built LLM agents in production, not demos: harness structure, prompt architecture, tool calling, RAG, MCP, context management, and debugging pipelines that fail differently every time. This is the core of the role.
• Evaluation discipline — you treat "did this change help?" as an engineering question with an answer, not a vibe check.
• Strong backend engineering in TypeScript/Node, plus working Python. You own your features all the way to the screen, so you need to build a competent frontend without waiting for someone else.
• Fluency building with agents — you use Claude Code or an equivalent harness as a real force multiplier: skills, subagents, hooks, custom tooling. And what comes out the other end is a deliberate, secure, maintainable feature, not the volume of half-understood code these tools make it so easy to generate. You own every line you ship, whether or not you typed it.
• Autonomy — you take a scope rather than instructions and carry it to done, including the parts nobody wrote down.
• Security instinct — the habit of asking "how would someone break this" before shipping.
• Taste in interfaces, and we mean something specific: sufficient minimalism. In the age of AI it is trivially easy to build a Christmas tree that looks impressive and solves nothing.
• Working proficiency in Russian and English. Our engineering teams work in Russian; documentation and company-wide collaboration are in English.

Nice To Haves

• a background in financial services, brokerage or fintech
• Slack apps, CRM integrations, OAuth / on-behalf-of flows
• agent observability and tracing at scale

Responsibilities

• Build and own the agent layer: harness design, prompt architecture, the tool layer, orchestration and context strategy.
• Build and own evaluations — the mechanism that tells us a change made the agent better rather than merely different.
• Build the backend behind it: APIs, services, data models, integrations with company systems — plus the frontend that exposes them.
• Design with an attacker in mind: prompt injection, cross-tenant data leakage, access control. We handle client financial data.
• Extend the scope where it needs it — edge cases, failure modes, UX, and the metric that proves it worked.
• Dig into unfamiliar infrastructure, find the right people, get to a decision.

Benefits

• Competitive salary that reflects your experience and the value you bring.
• Flexibility that fits your life — work from home, from our office, or a mix of both. You decide what works best.
• Flexible benefits package — choose the options that suit your life, not a one-size-fits-all bundle.
• A genuinely good place to work — an informal, collaborative culture where ideas are heard and bureaucracy stays out of your way.
• Continuous learning — ongoing training, education programs, and the support to deepen your expertise in a fast-moving industry.
• Connection beyond your desk — events that bring our teams together to network and celebrate.
• Global exposure — work side by side with talented colleagues from all over the world, across a business serving clients in 100+ countries.

What This Job Offers

• Job Type: Full-time
• Career Level: Mid Level
• Education Level: No Education Listed