Operating ML pipelines and containerized workloads (Airflow, Docker, Kubernetes)
Managing model lifecycle in production (deployment/serving: ONNX, TorchServe, Triton; registry/versioning: MLflow)
Monitoring, observability, data validation for production systems (Prometheus, Grafana)
Trách nhiệm
Own continuous training, validation, deployment pipelines with rollback safety
Build end-to-end observability for data, features, models, serving; lead incident response
Partner with Backend on serving performance, stability, cost-efficiency at scale
Collaborate on centralized feature store design and rollout across training and inference
Có thì tốt
Experience with feature stores (Feast, Tecton, Hopsworks) or building centralized feature pipelines
Java or Scala — online serving runs on the JVM
Streaming and low-latency stores: Kafka, Redis
Lương và phúc lợi
Lương tháng 13
Tài khoản AI (Claude) do công ty cung cấp
Môi trường làm việc tự chủ, tác động trực tiếp đến doanh thu quảng cáo
Tóm tắt bằng AI từ JD gốc, đọc JD gốc trước khi ứng tuyển.
ML Lifecycle Reliability
Own the pipelines that continuously train, validate, and deploy models, so only models that meet quality standards reach production — and can be rolled back safely when they don't
Make every production model versioned, traceable, and reproducible, and set the engineering standards that let new models move from experiment to production smoothly.
Observability
Build end-to-end visibility into data, features, models, and serving, so issues are detected and resolved before they impact the business
Lead ML-related incident response and drive root-cause fixes that prevent recurrence.
Serving Performance
Partner with Backend to keep model serving fast, stable, and cost-efficient as traffic and model complexity grow.
Feature Platform
Collaborate with Data Science, Data Engineering, and Backend on the design and rollout of a centralized feature store, so features are defined once and served consistently across training and inference.
Technical Skills
3+ years in ML engineering, MLOps, or backend/data engineering supporting ML systems in production
Strong Python and software engineering practices — testing, code review, Git workflows, and CI/CD
Operating ML pipelines and containerized workloads (Airflow or similar; Docker, Kubernetes)
Managing the model lifecycle in production — deployment and serving (ONNX, TorchServe, Triton, or similar), model registry and versioning (MLflow or similar)
Monitoring, observability, and data validation for production systems (Prometheus, Grafana, or similar)
Working knowledge of deep learning (PyTorch) and large-scale data processing (Spark, SQL, or similar)
Strong debugging and systems problem-solving — able to investigate failures across data pipelines, training, deployment, and inference.
Soft Skills
Strong ownership and reliability mindset
Communicates clearly across Data Science, Backend, and Data Engineering.
Nice-to-Have Qualifications
Experience with feature stores (Feast, Tecton, Hopsworks) or building centralized feature pipelines
Java or Scala — our online serving runs on the JVM
Streaming and low-latency stores: Kafka, Redis
Familiarity with ClickHouse or other OLAP databases
GPU inference optimization, vector search, or approximate nearest neighbor retrieval
Ad tech or recommendation system domain experience
An AI-first mindset — using AI/LLM tools to automate operations and accelerate engineering.
Why Join Adtima?
Large-scale ML challenges: run real-time ML systems for an advertising ecosystem serving over 80 million users and hundreds of millions of ad impressions daily
Technical and business exposure: hands-on experience across ML infrastructure, model serving, data pipelines, and advertising systems
Meaningful impact: the reliability and performance of the ML system directly protect ad revenue and advertiser trust
End-to-end ownership: shape the ML platform — from feature store to deployment and monitoring — with a high degree of autonomy
AI-powered workflow: company-provided AI accounts (Claude) to boost your productivity in engineering and operations
Large-scale ML challenges: run real-time ML systems for an advertising ecosystem serving over 80 million users and hundreds of millions of ad impressions daily
Technical and business exposure: hands-on experience across ML infrastructure, model serving, data pipelines, and advertising systems
Meaningful impact: the reliability and performance of the ML system directly protect ad revenue and advertiser trust
End-to-end ownership: shape the ML platform — from feature store to deployment and monitoring — with a high degree of autonomy
AI-powered workflow: company-provided AI accounts (Claude) to boost your productivity in engineering and operations.
Adtima is the leading advertising platform in Vietnam's Zalo ecosystem, matching ads to more than 80M users in real time and serving hundreds of millions of ad impressions every day. We are looking for a Senior Machine Learning Engineer to own the production side of our ads recommendation system. Our models retrain continuously and serve in real time; your mission is to make sure the whole ML lifecycle — data, features, training, deployment, and serving — runs smoothly, reliably, and transparently. You will build the platform and standards that let our Data Scientists ship new models quickly and safely, working closely with Backend and Data Engineering. This role is focused on the reliability and performance of the ML system. Model development and experimentation are owned by our Data Scientists, so you can concentrate on making the ML platform production-grade.
Adtima is the leading advertising platform in Vietnam's Zalo ecosystem, matching ads to more than 80M users in real time and serving hundreds of millions of ad impressions every day. We are looking for a Senior Machin...
