Forward Deployed Engineer - Agentic AI - Hybrid

Genesis10•New York, NY

6d•Onsite

About The Position

Genesis10 is seeking a Forward Deployed Engineer (FDE) – Agentic AI Workflow Automation for a 5-month contract position with a financial services client. This role involves embedding directly within one or more internal business functions to understand their workflows firsthand, and using agentic AI models to design, build, and iterate automation solutions. The FDE will own the full lifecycle within their embedded function, from discovery to iteration, while feeding reusable patterns back to the broader engineering organization. This is a hands-on, high-autonomy role that blends engineering delivery with direct business partnership, requiring the FDE to help define and prioritize automation opportunities.

Requirements

Total experience: 8–15+ years in software engineering, with meaningful recent experience building and shipping production AI/ML or agentic systems (not pilots or proofs of concept).
Forward-deployed / embedded delivery experience: Prior experience working directly embedded with a business function or client — as a forward-deployed engineer, solutions engineer, or similarly hands-on customer-facing technical role — rather than purely centralized platform engineering.
Domain background: Prior experience in a large regulated enterprise — banking, financial services, insurance, or healthcare — where AI systems had to satisfy audit, security, and compliance review before going live.
Process fluency: Demonstrated ability to sit with non-technical business users, learn a messy manual process firsthand, and translate it into an automatable workflow without extensive requirements documentation.
Autonomy: Comfortable operating with significant independence inside a division, making judgment calls on scope and design without constant central oversight.
Strong proficiency in Java and/or Python for building and shipping production tools quickly.
Strong SQL and data-modeling skills; comfort working against legacy enterprise data stores.
Hands-on experience using agentic AI models and frameworks (tool-calling / function-calling, multi-step orchestration) to build working solutions quickly, rather than building agentic infrastructure from scratch.
Retrieval-augmented generation for unstructured data (contracts, policy documents, correspondence, reports) common in control, risk, and compliance-adjacent workflows.
Comfort working with vector databases and embedding pipelines at a practical, applied level (e.g., Milvus, Weaviate, Pinecone).
Ability to integrate quickly with a division's existing systems, data stores, and internal APIs without lengthy platform build-out.
Cloud deployment experience, ideally AWS.
Comfort standing up lightweight monitoring for solution accuracy, usage, and reliability in a fast-moving embedded environment.
Working knowledge of human-in-the-loop design patterns and audit-by-design logging suitable for a regulated decision-making environment.
Enough fluency in model evaluation to flag when a workflow is not a good fit for automation, or needs additional guardrails.

Nice To Haves

Prior experience as a Forward Deployed Engineer or in a similarly embedded delivery model at a technology company known for that approach is a strong plus.
Prior work automating workflows in internal control, risk, compliance, financial reporting, or shared-services functions specifically, rather than only customer- or trading-facing systems.
Experience using autonomous coding or task agents as part of a fast-moving delivery workflow.
Exposure to financial industry regulatory standards and how they constrain automation design.
Demonstrated comfort being the sole embedded technical resource within a business team, without a large supporting engineering team on-site.

Responsibilities

Embed with assigned business function(s) to learn their workflows firsthand — shadowing end users, reviewing existing processes, and identifying high-value automation candidates in partnership with divisional leadership.
Design and build agentic AI solutions using existing agentic frameworks and foundation models, favoring rapid prototyping and fast iteration with real users over long design cycles.
Own delivery end to end within the embedded function: discovery, prototyping, pilot, production deployment, and ongoing iteration based on user feedback.
Translate ambiguous, often undocumented manual processes — as described by non-technical business users — into clear automatable workflows and working solutions.
Apply human-in-the-loop controls and structured, auditable logging so every automated decision is explainable and defensible to regulators and internal audit.
Continuously gather feedback directly from divisional end users and adjust solutions in near-real-time rather than through a formal ticketing/backlog process.
Feed successful patterns, reusable components, and lessons learned back to the central engineering organization so other FDEs and functions can reuse them.
Partner with security and compliance stakeholders as needed to clear solutions for production use within the function.