Zscaler
About this role
Zscaler is a cloud security company. Its Zero Trust Exchange sits between employees and the applications they use, inspecting traffic so attacks and data loss are caught before they land, and the company describes running the world's largest security data lake behind it. This role is on the Exposure Management and Security Operations team, working on a multitenant platform the posting says serves over 15 million users globally. The job is applying agentic AI to that problem: building AI agents and getting them all the way into production, which the posting is refreshingly specific about, naming multi step reasoning, tool use, state management and language model orchestration in live environments as the hard parts. You would also build the LLMOps layer underneath, meaning the observability, monitoring, logging and evaluation that dynamic agents need, and keep improving both the newer generative pipelines and the existing classical models. At 3 to 5 years it is one of the more reachable AI roles in today's edition.
Who this is for
What the posting requires
- 3 to 5 years as a machine learning engineer, with a track record of building and deploying ML systems
- Hands on experience creating generative AI systems including large language models and retrieval augmented generation, plus a clear history of operationalising agentic AI in production
- Deep understanding of MLOps and LLMOps, specifically the monitoring, observability and debugging problems created by non deterministic agents
- A solid computer science foundation in data structures and algorithms
- Python expertise with scikit-learn and PyTorch or TensorFlow, plus SQL for feature engineering and model evaluation
- Strong communication, able to work independently and in a team
Counts as a bonus, not a requirement
- Cloud services such as AWS, GCP or Azure, and ML platforms such as Kubeflow or SageMaker
- Familiarity with systems programming or distributed systems
- Visible contribution to the field such as technical blogs, papers or industry knowledge sharing
What you would actually be doing
- Independently developing machine learning models and generative AI solutions, with the emphasis on building and deploying agentic architectures
- Driving AI agents end to end into production, handling multi step reasoning, tool use, state management and model orchestration in live environments
- Designing LLMOps frameworks that give real observability, monitoring, logging and evaluation for agents
- Refining and optimising both generative pipelines and existing classical models through feature engineering and hyperparameter tuning
- Working with cross functional teams to turn business needs into technical solutions
Location and working pattern
A hybrid position based in Bangalore, reporting to the Manager of Machine Learning Engineering.
A good fit if
You have shipped machine learning before and want to move into agent based systems, on a security product where a mistake has real consequences. Security data is adversarial in a way most recommendation and forecasting work is not, because someone is actively trying to defeat your model.
Think twice if
You have only used language models through an API. The posting asks specifically for agents you have run in production, and repeats the point about non deterministic behaviour, which is where that experience shows.
- 3 to 5 years as a machine learning engineer, with a track record of building and deploying ML systems
- Hands on experience creating generative AI systems including large language models and retrieval augmented generation, plus a clear history of operationalising agentic AI in production
- Deep understanding of MLOps and LLMOps, specifically the monitoring, observability and debugging problems created by non deterministic agents
- A solid computer science foundation in data structures and algorithms
- Python expertise with scikit-learn and PyTorch or TensorFlow, plus SQL for feature engineering and model evaluation
- Strong communication, able to work independently and in a team
Counts as a bonus, not a requirement
- Cloud services such as AWS, GCP or Azure, and ML platforms such as Kubeflow or SageMaker
- Familiarity with systems programming or distributed systems
- Visible contribution to the field such as technical blogs, papers or industry knowledge sharing
What you would actually be doing
- Independently developing machine learning models and generative AI solutions, with the emphasis on building and deploying agentic architectures
- Driving AI agents end to end into production, handling multi step reasoning, tool use, state management and model orchestration in live environments
- Designing LLMOps frameworks that give real observability, monitoring, logging and evaluation for agents
- Refining and optimising both generative pipelines and existing classical models through feature engineering and hyperparameter tuning
- Working with cross functional teams to turn business needs into technical solutions
Location and working pattern
A hybrid position based in Bangalore, reporting to the Manager of Machine Learning Engineering.
A good fit if
You have shipped machine learning before and want to move into agent based systems, on a security product where a mistake has real consequences. Security data is adversarial in a way most recommendation and forecasting work is not, because someone is actively trying to defeat your model.
Think twice if
You have only used language models through an API. The posting asks specifically for agents you have run in production, and repeats the point about non deterministic behaviour, which is where that experience shows.
Apply on company site
Opens job-boards.greenhouse.io, the employer's own application page. Applying is always free.