Senior Backend and Agent Engineer

Ambient Skye
· Remote, based in Connecticut
· Senior / Staff

Reports to:

Director of Software Development

The Senior Backend and Agent Engineer builds and co-owns the AI core of Ambient Skye's strategic-intelligence platform: the agent orchestration, the actor and scenario engines, and the report pipeline that turn a hard question into findings traceable to every source, assumption and model call behind them. The role is the second owner of this code alongside the platform architect. Its work carries across every application the platform serves — scenario analysis and wargaming for government analysts, commercial intelligence for life-sciences companies, and investment analysis — so the requirements below are the same whichever application the role is working on.

• Own the reliability of the LLM call path: the tool-use loop, context limits, provider failover, structured output, cost and latency.
• Extend the agent orchestration and the multi-perspective challenge engines that test every finding before it reaches a customer.
• Keep prompts and model configuration versioned, so any output can be traced to the model and prompt that produced it.

• Extend the actor and scenario engines: structured, editable actor profiles, move and counter-move between AI actors and human participants, analyst overrides and event injection.
• Implement branching and replay — a fork preserves the original, a replay uses pinned model versions and recorded evidence, and every move is attributed to the agent or person who made it.
• Connect the quantitative scoring engine to scenario outcomes.
• Translate product requirements into technical designs and well-scoped, deliverable work across the layered backend (routes, services, helpers, database).
• Review a high volume of AI-assisted code with a critical eye; design judgment, not typing speed, is the bottleneck.
• Work within the engineering quality gates: tests, coverage, linting and security scans.

Ambient Skye is a small, tightly collaborative team; this role succeeds through partnership, not silos.

• Works most closely with the Director of Software Development on day-to-day priorities, technical design, code review and delivery commitments.
• Partners with the COO, who is also the platform architect, on the design of the AI core, the scenario engines and the scoring method.
• Works with the VP of User Experience so analyst-facing behaviour (branching, overrides, traceability) is designed together with the screens that expose it.
• Communicates technical context clearly to non-engineering partners and customers; some deliverables go directly to analysts and executives.

• 7+ years building production backend systems in Python. Flask or a similar framework is fine; we use Flask-Smorest and Marshmallow.
• Production experience building LLM agents with tool use: has shipped a loop that calls tools and knows how it fails.
• Strong PostgreSQL and hand-written SQL. We use no ORM, and every query runs under row-level security for tenant isolation.
• Experience modelling versioned or event-sourced state (forks, replays, audit trails) and reasoning about determinism.
• Comfort working inside strict CI gates and a layered architecture.
• Clear written communication.
• US person (for example, a US citizen or permanent resident).

• Simulation, scenario-modelling, game-engine or multi-agent systems experience.
• Experience with the Anthropic or OpenAI APIs, structured output and prompt versioning.
• Eligibility for a security clearance (not required).

• Builds tenant data isolation (row-level security) into every query, agent tool and data path, and treats a cross-tenant leak as the highest-severity defect.
• Emits audit events for every state-changing operation and never suppresses an audit failure.
• Follows the platform's change-management controls: reviewed pull requests, CI gates and approved deployments.
• Keeps customer-supplied material and model inputs and outputs within the platform's access, encryption and retention controls.
• Supports security incident response for the components the role owns, and triages dependency and vulnerability findings on the defined timeline.

• Base salary $160,000 – $190,000, plus equity.