• Build a learned world model of the welding process that predicts future system behavior under robot actions. • Develop multimodal neural simulators incorporating signals such as 3D scans, video, thermal data, and electrical measurements. • Design, train, and evaluate large-scale generative or dynamics models (e.g., video prediction, latent world models, 3D or spatiotemporal representations) capable of long-horizon rollouts. • Collaborate with reinforcement learning engineers by integrating the neural simulator into RL pipelines for policy training and evaluation. • Run research tracks in parallel with production development, including hypothesis-driven experimentation and ablation. • Partner closely with data and MLOps teams to support scalable training, evaluation, and deployment - while remaining comfortable owning pieces of the stack when needed. • Translate research prototypes into robust, maintainable production code when they prove valuable. • Validate simulator performance against real-world robotic welding data and support sim-to-real transfer.

• Design and implement comprehensive evaluation frameworks that reflect real-world task success for agentic systems, with a focus on human+AI collaboration outcomes • Build benchmarking pipelines that capture nuanced success indicators including trust calibration, intervention frequency, and agent handoff quality • Lead development of observability tools and instrumentation for analyzing agent behavior in production • Translate complex qualitative and quantitative signals into actionable insights that inform model iteration and product prioritization • Collaborate with researchers, engineers, and product teams to align evaluation methodologies with business and user goals • Own benchmarking infrastructure that enables reproducible, scalable evaluation across AI initiatives • Champion rigorous experimental design and statistical analysis across teams to ensure consistent and meaningful measurement standards

• Design and train advanced ML models (e.g., DNNs, transformers) to power Reddit Ads Ranking models. • Develop and optimize features including embeddings, contextual signals, and cross-session behavior. • Own experimentation strategy: offline metrics, A/B design, readouts, and “next iteration” planning. • Work closely with product, infra, and data teams to drive end-to-end model deployment and performance analysis. • Mentor other MLEs and contribute to modeling best practices across the org. • Shape the long-term modeling vision across one or more domains (conversion, app ads, shopping, brand, etc.).

• Architect and significantly influence ML training/serving systems and tooling that unlock faster iteration and larger models. • Drive adoption and reliability of ML systems and own a portfolio of initiatives across multiple teams. • Improve efficiency: GPU utilization, training runtime, data loading, feature performance, and serving latency. • Write high-quality design docs; run design reviews; set standards for correctness, reliability, and velocity. • Partner cross-org on shared infra sequencing, requirements, and adoption. • Mentor other engineers and contribute to engineering best practices across the org.