AI & Data Science Director

Reflex Media, Inc.•, (Multiple States)

21h•$175,000 - $225,000•Remote

About The Position

Seeking.com is the world's largest premium dating platform, founded and led by an MIT alumnus and headquartered in Las Vegas. We are a lean company of about 60 people, so every leader here builds. You will lead our combined Data Engineering & Analytics, AI, and Data Science team of six to seven individual contributors and report directly to the Co-CEOs. This team is Seeking's central AI capability: the models, infrastructure, evaluation frameworks, and continuous-learning systems the whole company runs on. Expect about half your time leading and half writing code, building models, and debugging production systems yourself. Your first mandate is to achieve 100% AI automation of routine user-uploaded content moderation, with accuracy proven superior to our human moderators and at a lower total cost. Next is AI customer support that resolves member tickets end to end. You own both outcomes completely, from policy and data through production results. Over time, the team will extend AI to fraud prevention, matching, member onboarding and success, marketing, monetization, retention, and product development, in partnership with the leaders who own those areas. Every initiative carries a measurable target, a named owner, and a mechanism for continuous learning. The immediate problems come first, and this role exists to solve them. This role is for someone who manages a hands-on team today and still ships code. If your recent roles had managers reporting to you, or you plan to delegate the technical work and manage through status meetings, this is the wrong role. If you walk into ambiguity and know exactly what to do, keep reading.

Requirements

A Bachelor’s in Data Science, Computer Science, Software Engineering, or a closely related technical field.
Seven or more years delivering production data or ML systems, including three or more years managing teams of individual contributors.
You manage a hands-on team today, or did so recently.
Strong Python and SQL, with production code you personally wrote in the past year.
Production ML and architecture judgment. You have designed, built, and operated ML and LLM systems end to end, from data pipelines through evaluation, deployment, and monitoring, on a major cloud platform.
You have done this before. You led an AI automation program in moderation, trust and safety, support, or a comparable decision system, from a measured human baseline to automated decisions at scale.
Evaluation rigor and continuous learning. You measure false negatives and false positives separately, report uncertainty, keep test data clean, and have built systems that improve through feedback and controlled updates.
Measurable business impact. You can name the cost, revenue, efficiency, or member outcomes your work moved, with the numbers.
Cross-functional influence. You have found AI opportunities in other departments, won over their leaders, and taken the solution from concept into production.
A record of mentoring engineers and scientists into stronger, more senior contributors.
Daily use of Claude, Claude Code, Cowork, or comparable AI tools. If someone has to explain to you why AI matters, this role is not for you.

Nice To Haves

Production work in multi-modal or computer vision moderation, fraud detection, recommender systems, or marketing and monetization models such as propensity, pricing, and lifetime value.
Our stack: AWS (S3, Redshift, SageMaker, Bedrock), Looker, and Kubernetes.

Responsibilities

Set the foundation in your first 30 days: Bring Product, Support, and Operations to one written moderation policy with a single owner. When teams disagree, you run the process that settles it. Inventory every production AI system with a named owner, backup, dependencies, and monitoring gaps. Trace every upload from receipt through model decision, enforcement, and any human review. Quantify every path that still needs a human, with its root cause, cost, and fix.
Prove AI beats our human moderators, then reach 100%: Test the AI and our moderators on the same held-out cases, with reference decisions set by independent, policy-qualified reviewers. Report missed violations and wrongful rejections separately, with sample sizes and confidence intervals. Close every gap: missing coverage, uncertain model outputs, system failures, unenforced decisions, and duplicate human review. Expand automation in controlled stages, each with a tested rollback.
Build systems that keep getting better: Every production system improves through verified feedback, independent evaluation, and controlled releases. Run a monthly learning cycle in which a new model ships only when it beats production on the same independent test. Version every dataset, policy, prompt, and model so every gain can be proven and every change can be rolled back.
Run production AI like a business: Give every production AI system monitoring, alerts to a named owner, documented recovery, and regular failure drills. Report results in business terms: quality, cost per outcome, revenue, efficiency, and member experience, with savings reconciled with Finance. Keep one executive dashboard and run a weekly review with the Co-CEOs.
Automate support, then partner across Seeking: Own AI customer support end to end: agents that resolve tickets and execute approved actions, such as subscription changes and eligible refunds, held to the same evaluation, rollout, and monitoring standards as moderation. Once moderation and support are delivered, partner with the leaders of Marketing, Product, and other departments on fraud, matching, onboarding, monetization, and retention. Bring the Co-CEOs a ranked list of opportunities by impact on cost, revenue, efficiency, and member experience. In those partnerships, department leaders own their business outcomes. Your team owns the models, infrastructure, evaluation, and learning systems, and is accountable for their performance.
Lead and grow the team: Review each person's work line by line, pair on hard problems, and give direct feedback every week. Build a growth plan for everyone, recruit strong talent, and address performance gaps early. Set quarterly OKRs for the function and each direct report, with leading and lagging KPIs for every Key Result. Run the work in Jira, Confluence, and Monday.