Forward Deployed Engineer

Space Executive; New York (NY); On-site; USD 150,000 - 210,000; Full time

Job summary

Space Executive in New York is seeking an experienced professional to own the end-to-end AI agent lifecycle, from design to deployment and optimization. You will work across the stack—client, backend, agent runtime, tool layer, and real-time streaming—using Python or TypeScript.

The role requires deploying LLM workflows ( LangGraph ) and agents ( Claude Code SDK ) at production scale, with a focus on async design and distributed systems.

Qualifications

• End-to-end AI agent experience in production environments.
• Experience deploying LLM workflows and agents at scale.
• Strong grasp of async/streaming design and distributed systems.
• Ability to form and verify hypotheses with limited data.
• Fluent English communication with global colleagues.

Responsibilities

• Turn ambiguous requirements into concrete problems from design to implementation.
• Define core versus edge AI agent behaviors for customers.
• Operate agent behavior as configurable prompts and routing.
• Build production infrastructure, APIs, data pipelines, and runtime env.
• Develop evaluation and testing systems with measurable improvements.
• Trace and debug distributed systems end-to-end.

Skills

AI agent design

Python/TypeScript

Production deployment

LangGraph experience

Claude Code SDK

Async/Streaming

Distributed systems

English fluency

Tools

kubectl

Helm

ArgoCD

KEDA/HPA

Terraform

AWS/Azure/GCP

Job description

New York · Full-time · All levels (new grad through very senior)

Who They Are

The Company is a New York-based, AI-native company building the Alpha Intelligence Layer for global public markets. Founded by a team of former investment bank analysts and MIT computer science PhDs, the company recently closed a Series A led by a syndicate of strategic financial institutions and venture platforms across the US, Europe, and Asia.

More than 70 financial institutions across the US, Europe, and Asia use the platform every day for real research work — single-name analysis, earnings and disclosure interpretation, investment due diligence, and market briefings. That includes sell-side sales, trading, and research teams at leading investment banks, and buy-side clients collectively managing more than $5 trillion in assets.

The team is globally distributed, with core hubs in New York and Seoul, plus members in the UK, Singapore, and Hong Kong. They work closely with in-house finance-domain experts, including former buy-side and sell-side analysts.

The Role

The FDE owns the last mile between the Company's core agent product and the reality of how each institutional customer actually works. Rather than plain systems integration, this is about defining each customer's real problem deeply enough to see their specific pain points — while generalizing across customers to decide what belongs in the core agent versus what should be solved at the edge.

Agent behaviour (prompts, tools, models, routing) is version-controlled configuration rather than hard-coded logic, assembled and deployed without a code push. The role spans the full stack: client → backend services → the agent runtime → the tool layer (MCP) → real-time streaming (SSE) — and requires being able to pinpoint where latency, bottlenecks, or failures arise anywhere along that path.

What You'll Do

• Turn ambiguous customer requirements into concrete, solvable problems — owning the work from design through implementation and improvement.
• Design the boundary between core agent behaviour and customer-specific workflows (market-briefing automation, DD-report agents, document search/citation, natural-language querying over structured data).
• Operate agent behaviour as configuration — versioning prompts, tools, models and routing through a draft → simulation → deploy cycle.
• Build production infrastructure yourself — API integrations, connectors, data pipelines, and the execution environment the agent runs on (nodes/Pods, autoscaling).
• Build evaluation and test systems (LLM-as-judge, quality scores, citation/source-grounding checks) and ship measurable improvements from real usage logs and traces.
• Trace and debug distributed systems end to end across multiple services and data stores.
• Work daily in English across Business, Product, engineering, infrastructure, and in-house finance-domain experts.
• Use coding agents (Claude Code, Cursor) as a core part of the daily workflow.

What They're Looking For

• End-to-end experience designing, deploying, monitoring, and improving AI agents in production, in Python or TypeScript — not just prototypes.
• Experience deploying LLM workflows (e.g. LangGraph) and agents (e.g. Claude Code SDK) at real production scale.
• Strong grasp of async/streaming design, distributed systems, and where to draw context boundaries.
• Comfortable with ambiguity — able to form and verify hypotheses quickly with limited data.
• Fluent, confident English communication — reading dense technical/financial material and working directly with global colleagues and customers.
• AI-native: fluent with coding agents such as Claude Code and Codex.

Nice to Have

• Background in micro/macroeconomics or hands-on experience in a finance domain (buy/sell side, front/mid/back office).
• Experience in forward-deployed engineering, technical consulting, or other customer-facing technical work.
• Experience building LLM evaluation systems (LLM-as-judge, quality scoring, citation/source grounding).
• Hands-on Kubernetes/GitOps (kubectl, Helm, ArgoCD, KEDA/HPA) and IaC (Terraform).
• Cloud experience (AWS/Azure/GCP) and production observability (e.g. Datadog).