Sr. ML Engineer

Role Description

Turn audience science into scalable production systems that help advertisers decide whom to reach. Working with terabyte-scale data, you'll build, score, refresh, and deliver audience segments to activation endpoints across DSPs, clean rooms, CTV, social, and programmatic partners. You bridge data science and platform engineering—converting prototypes into production-grade services and owning the pipelines, MLOps, and integrations behind them. As a senior team member, you'll set engineering standards, lead end-to-end system design, and mentor other engineers. Stack: Databricks, Spark, Snowflake, and Azure.

Responsibilities

• Convert data science prototypes (propensity, lookalike, segmentation, fusion) into reproducible, production-quality ML services.
• Build and operate large-scale data/feature pipelines on Databricks + Spark, reading from and writing to Snowflake, producing versioned audience segments on schedule.
• Own the full model lifecycle: MLflow tracking/registry, CI/CD (Azure DevOps or GitHub Actions), automated retraining, drift monitoring, and rollback.
• Engineer the activation layer, governed delivery of segments and scores into DSPs/SSPs, DMP/CDP, clean rooms, and CTV/social/programmatic partners, including identity resolution and onboarding.
• Ensure scalability, latency, cost-efficiency, and reproducibility across Azure (ADLS, Azure ML, AKS); support batch and near-real-time scoring.
• Enforce privacy-by-design: data minimization, access controls, encryption, and clean-room-compatible patterns.

Qualifications

• 8–9 years hands-on ML engineering, shipping and operating production systems at scale.
• Prior experience in media / advertising / audience activation.
• Databricks (incl. MLflow and Spark), Snowflake, and production Azure (ADLS, Azure ML, Azure DevOps, AKS).
• Expert Python and ML ecosystem; strong SQL at scale.
• Strong applied statistics (sampling/weighting, forecasting, causal inference, or experimental design).
• CS/engineering degree with strong quantitative foundations.

Requirements

• Agile practices, Docker/Kubernetes, advanced Databricks/Snowflake (Structured Streaming, tuning, Model Serving/Feature Store), orchestration (Databricks Workflows, Airflow, Dagster).
• Ad tech/identity experience: DSP/SSP, DMP/CDP, clean rooms (LiveRamp, Snowflake), identity graphs, onboarding, delivery APIs.
• AWS/GCP a plus.

About you

• Pragmatic and deeply hands-on about production trade-offs; a clear communicator with non-expert stakeholders; a proactive team player who mentors others.
• Comfortable working the second half of the day to overlap with a US-based product team.
• Fluent in an English-language business environment.

Company Description

Fusemachines is a leading AI strategy, talent, and education services provider. Founded by Sameer Maskey Ph.D., Adjunct Associate Professor at Columbia University, Fusemachines has a core mission of democratizing AI. With a presence in 4 countries (Nepal, the United States, Canada, and the Dominican Republic) and more than 450 full-time employees, Fusemachines brings global AI expertise to transform companies worldwide. Founded in 2013, Fusemachines is a global provider of enterprise AI products and services, on a mission to democratize AI. Leveraging proprietary AI Studio and AI Engines, the company helps drive the clients’ AI Enterprise Transformation, regardless of where they are in their Digital AI journeys. With offices in North America, Asia, and Latin America, Fusemachines provides a suite of enterprise AI offerings and specialty services that allow organizations of any size to implement and scale AI. Fusemachines serves companies in industries such as retail, manufacturing, and government.