Director of AIML in our AdTech team, you'll drive innovation in our advertising platform using advanced machine learning and statistical methods. You'll tackle complex problems in real-time bidding, ad ranking optimization, budget efficiency, and performance measurement that directly impact our business's bottom line. This role requires technical excellence and business acumen, and you'll collaborate with cross-functional teams to transform data insights into actionable strategies that enhance our advertising ecosystem.

Requirements:

Education: Bachelor's or master's degree in computer science, engineering, or a related field.
Experience: 15+ years in software/ML engineering, with 8+ years building production ML systems and 5+ years in senior technical leadership, ideally in high-scale consumer internet or e-commerce.
Proven experience building and leading an MLE function from a nascent starting point and partnering with applied DS teams as a production-engineering counterpart, not a replacement.
Deep hands-on background in online ML serving, feature stores, and experimentation platforms at consumer scale, and a track record of building and retaining technical teams in high-attrition-risk environments.

ML Serving and Production Systems (Core):

Low-latency online inference: model servers (Triton, KServe, Ray Serve, or equivalents), batching, model optimization (quantization, ONNX/TensorRT), autoscaling, and p99 latency engineering.
Feature store architecture (Feast or equivalents): online/offline consistency, point-in-time correctness, streaming features, skew prevention.
Training and pipeline infrastructure: distributed training on Ray or equivalent, PyTorch/XGBoost, orchestration (Argo, Airflow, Kubeflow), experiment tracking and model registry (MLflow or equivalent), shadow/canary rollout, CI/CD for ML, and production model observability (drift detection, data validation, alerting).
Strong software engineering fundamentals: Python plus a systems language (Java/Go/C++), gRPC service design, Kafka/Flink streaming, Redis-class caching, Kubernetes-native deployment, and infrastructure-as-code.

Personalization, Retrieval, and Ranking Domain:

Shipped two-tower/dual-encoder retrieval and embedding-based candidate generation at consumer scale, including embedding pipelines and index refresh/freshness trade-offs.
Deep familiarity with ANN retrieval and vector databases (Qdrant, Vespa, Milvus, HNSW/ScaNN internals) and hybrid lexical + vector retrieval, including production operations (sharding, compaction, tail latency).
Multi-stage ranking design under strict latency budgets and experimentation literacy: A/B testing at scale, interleaving, and logging design for unbiased evaluation.

GenAI Production Engineering:

Requirements:

Education: Bachelor's or master's degree in computer science, engineering, or a related field.
Experience: 15+ years in software/ML engineering, with 8+ years building production ML systems and 5+ years in senior technical leadership, ideally in high-scale consumer internet or e-commerce.
Proven experience building and leading an MLE function from a nascent starting point and partnering with applied DS teams as a production-engineering counterpart, not a replacement.
Deep hands-on background in online ML serving, feature stores, and experimentation platforms at consumer scale, and a track record of building and retaining technical teams in high-attrition-risk environments.

ML Serving and Production Systems (Core):

Low-latency online inference: model servers (Triton, KServe, Ray Serve, or equivalents), batching, model optimization (quantization, ONNX/TensorRT), autoscaling, and p99 latency engineering.
Feature store architecture (Feast or equivalents): online/offline consistency, point-in-time correctness, streaming features, skew prevention.
Training and pipeline infrastructure: distributed training on Ray or equivalent, PyTorch/XGBoost, orchestration (Argo, Airflow, Kubeflow), experiment tracking and model registry (MLflow or equivalent), shadow/canary rollout, CI/CD for ML, and production model observability (drift detection, data validation, alerting).
Strong software engineering fundamentals: Python plus a systems language (Java/Go/C++), gRPC service design, Kafka/Flink streaming, Redis-class caching, Kubernetes-native deployment, and infrastructure-as-code.

Personalization, Retrieval, and Ranking Domain:

Shipped two-tower/dual-encoder retrieval and embedding-based candidate generation at consumer scale, including embedding pipelines and index refresh/freshness trade-offs.
Deep familiarity with ANN retrieval and vector databases (Qdrant, Vespa, Milvus, HNSW/ScaNN internals) and hybrid lexical + vector retrieval, including production operations (sharding, compaction, tail latency).
Multi-stage ranking design under strict latency budgets and experimentation literacy: A/B testing at scale, interleaving, and logging design for unbiased evaluation.

Data Scientist

Skills

Must have

Full description

Requirements:

ML Serving and Production Systems (Core):

Personalization, Retrieval, and Ranking Domain:

GenAI Production Engineering:

Data Scientist

Skills

Must have

Full description

Requirements:

ML Serving and Production Systems (Core):

Personalization, Retrieval, and Ranking Domain:

GenAI Production Engineering:

Data and Platform Fluency:

Leadership and Communication:

Operational Excellence: