Cars24 : ML Engineer
Brief Description of position:
About the Role
This role is for you if you are passionate about machine learning, data engineering and DS in general, and are also familiar with the data science stack. It will require basic close collaboration with data scientists so a basic understanding of DS stack in python will be useful.
Following is a representative list of what you will be doing but not limited to what is mentioned.
What You Will Do
- Collaborate with data scientists to optimize ML models for high-throughput, low-latency use cases
- Engineer high-reliability, high-performance services for sophisticated ML-driven functionality
- Suggest and design deployment methods on cloud infrastructure (docker, cloud function etc)
- Design and build our reporting, data version control and model deployment stack to help data scientists productionize their models and features faster
- Work to define Data version control, Model version control and maintenance
- Build and maintain DS/BI warehouse that serve internal tools and interfaces to improve the productivity of the team and improve the accessibility of our products
- Productionizing models is the primary task, though you would get to work on model building later
- You would serve as the bridge between the data scientists and software engineers.
- Enable efficient deployment, monitoring and debugging of production ML models
- Work with data scientists to make production code more robust in terms of scalability, latency and compute
Key skills required
- Understanding of optimizing code and writing high performance methods.
- Must have:
- knowledge of version control, proficiency in Python, deployment related best practices and enthusiasm towards MLOps
- Data science knowledge and familiarity with ML libraries such as Pandas, Scikit, Tensorflow, xgboost, Keras etc., will help in taking models to production and serving
- Good to have:
- Experience with web services and microservice architectures is a plus
- SQL/Big Data, MLOps knowledge, scaling ML models in high-throughput and low-latency settings
- GCP, Kubernetes/Docker knowledge, pipelining tools like Kubeflow
Maximum CTC (in lakhs per annum):