Staff AI Solutions Engineer at FanDuel in Atlanta, GA

Job description

THE POSITION Our roster has an opening with your name on it As a Staff AI Solutions Engineer at FanDuel, you will help us own the AI technical vision across the portfolio. Additionally, hold and evolve the technical vision across our governed agent layer, our strategy-optimization engine, and our simulation environment. In this role you will make the architectural calls that set the team’s direction and write the design docs and decision records that make those calls legible to everyone else. As well as, be the technical voice product that other teams trust. You will be expected to translate tradeoffs into decisions that they can act on and push back with evidence when a direction is wrong. If this describes you, read on – we want to hear from you! THE GAME PLAN Everyone on our team has a part to play Drive AI Into the Engineering Lifecycle Drive the integration of our agent layer into the data engineering SDLC: requests arriving from Slack, Jira, GitHub, and PagerDuty, routed into a planning phase that produces scoped, reviewable tickets and a build phase that carries each ticket through spec, plan, build, test, PR, and documentation. Build the environment that makes agentic engineering actually work: skills, specs, context scaffolding, shared memory, and progressive disclosure. Extend governed, audited data access so agents operate under each user’s own permissions against our data platform, with a limited blast radius by design. Define the outer loop: continual learning, post-run review, and agent self-improvement that lands as reviewable pull requests rather than uncontrolled drift. Make Optimization a Product Capability Advance our strategy-optimization engine from proof of concept to a capability product teams request by name, and land it against real business metrics. Partner with product to turn ambiguous business problems into optimizable ones: define the action space, the objective and utility weights that encode the tradeoffs we actually care about, and the guardrails and risk tolerance that keep the search honest. Own the deterministic scoring backbone: versioned scorers, shared and immutable experiment definitions, blind holdout discipline. Ensure results are comparable, auditable, and cumulative across the team rather than trapped in one person’s working directory. Expand the engine to new scoring and optimization domains across the business, including generosity and promotional spend, module placement and ranking, and search relevance. Bias toward interpretable strategies product partners can reason about, argue with, and implement without an external serving dependency. Prove Ideas Before They Ship Mature our offline simulation environment so product hypotheses can be pre-screened against synthetic personas built from real behavioral data, before committing scarce live A/B capacity. Establish directional agreement between simulated and live outcomes as the credibility bar, and be honest with partners about where the simulation is and is not trustworthy. Drive adoption of simulation as a standard input to experiment prioritization with our product partners. Turn Vision Into Work the Team Can Own Act as the Team’s technical lead: take strategic direction and decompose it into specs, skills, and tickets that individual engineers can own end to end without waiting on you. Set the working model for an AI-driven team. You define what a human specifies, what an agent builds, what a human reviews and raise the bar through code review, reusable patterns, and design review. Mentor engineers across the US and Romania pods, and write clearly, since not all collaboration is synchronous. Represent the team technically with partner engineering organizations, vendors, and external delivery partners on joint work. THE STATS What we’re looking for in our next teammate 8+ years of software, machine learning, or AI engineering experience, including technical leadership of a system other engineers built on. A track record of turnin