Reddit's Ads Creative Effectiveness team needs to build generative-AI and predictive products to help advertisers ship high-impact campaigns, aiming to raise the performance bar for every advertiser on the platform.
Requirements
- Deep ML Expertise: Hands-on experience with tree models, neural nets, ranking, and especially LLM/VLM fine-tuning & RLHF/RLPF for ads.
- Scalable Systems Know-How: Proven track record building high-throughput, low-latency services and real-time data pipelines.
- Quant & Experimentation Chops: Comfort with uplift metrics, variance, and causal inference ensures shipping impactful features.
- Prior work on ad-creative tooling (headline generators, pre-test models, auto-captioning, etc.).
- Experience navigating brand-safety, privacy, or copyright frameworks in generative AI products.
Responsibilities
- Ship and scale LLM/VLM-powered generators (images, video, text) and predictive models (creative quality scores, brand-lift).
- Own the full lifecycle from prototype to production.
- Architect low-latency generation/serving stacks and petabyte-scale feature pipelines that meet privacy, safety, and compliance requirements.
- Establish rigorous A/B and causal-inference frameworks; quantify lift, marketplace impact, and advertiser ROI.
- Integrate ACE capabilities into the self-serve flow and the ad auction.
- Embed brand-safety filters, copyright checks, and human-in-the-loop escalation paths to protect users and advertisers.
- Hire, mentor, and retain a diverse crew of MLEs and software engineers with a strong product mindset.
Other
- Lead & Grow: Hire, mentor, and retain a diverse crew of MLEs and software engineers with a strong product mindset.
- Foster a culture of rapid iteration, ownership, and inclusion.
- Set Vision: Work with product to define long term roadmaps that aligns generative AI bets with Ads OKRs.
- Translate fuzzy ideas into concrete milestones, measurable outcomes, and define technical steps that answer “how do we best build this?”.
- Partner Broadly: Collaborate with Product, Sales, Policy, UX, and sibling Ads ML teams to integrate ACE capabilities into the self-serve flow and the ad auction.
