Machine Learning Engineer · Bengaluru

Roles and Responsibilities

• ML Pipeline Architecture & EngineeringScale & Performance: Build and manage complex ML training and inference jobs using Databricks, ensuring high availability, fault tolerance, and data consistency.

• Pipeline Construction: Design and build robust, scalable ML pipelines (ETL, Feature Engineering, Training, Inference) using PySpark, SQL and Databricks

• Cloud-Native API Optimization: Implement model training code optimised for Google Cloud Platform (GCP), leveraging the specific nuances of Google Cloud Storage (GCS) and GKE to maximise throughput, and seamless CI CD. 

Model Deployment & Lifecycle Management

• Deployment Strategy: Operationalize ML models by building scalable inference services/APIs on Google Kubernetes Engine (GKE).

• Lifecycle Management: Implement MLOps best practices for model versioning, registry, and monitoring using MLflow (managed within Databricks or GCP). Code review and optimise datascience APIs for best in class practices.

• GenAI Operations: Deploy and manage Small Language Models (LLMs), Vision Language Models (VLMs), and other generative models using tools like Ollama or vLLM/TGI on GPU-accelerated infrastructure.

Optimization & Use Case Building

• Scale Optimization: Optimize PySpark jobs for performance and cost, tuning partitions, memory management, and caching or derived table strategies for massive datasets.

• Use Case Implementation: Partner with Data Scientists to take raw model prototypes and convert them into production-grade systems that solve specific business problems (e.g., Home Page Ranking, Search Ranking, Recommendations, GPU batching).

• Performance Tuning: Monitor and tune model latency and throughput, ensuring our deployments meet strict SLAs.

Desired Skills and Experience

• Experience: 2.5 to 5 years of hands-on experience in Machine Learning Engineering or Data Engineering with an ML focus.

Must Have

• ML Ops: Understanding of Architecture and Design of ML products. Be able to articulate the trade-offs.

• Cloud-Native DevOps: Experience with containerization (Docker) and orchestration (Kubernetes or ECS ) for ML services.

• Data for ML: Breadth of experience working with data blobs, delta lakes, SQL, and storage for ML workflows.

• GCP ML Deployment: Strong hands-on experience deploying models on Google Cloud Platform (Vertex AI or GKE).

• GCP Infrastructure: Familiarity with Kubernetes (GKE) and general GCP infrastructure (IAM, Networking concepts for ML).

• Databricks: Proven experience working with Databricks specifically within the GCP/Azure ecosystem (configuring clusters, Unity Catalog).

• PySpark: Deep understanding of writing, debugging, and optimizing PySpark jobs at scale.

• Core ML Ops: End-to-end experience with Model Training, Model deployment, Lifecycle Management, and Pipeline Orchestration.

Highly desired (Tech Stack Expansion)

• Strong proficiency in SQL for complex data analyses and extraction pipelines/optimisations. 

• Experience interacting with noSQL low latency databases such as Aerospike, Redis, MongoDB, ArangoDB.

• Ability to write optimised queries to support and put together A/B analytical dashboards with optimised derived table pipelines and queries

Good to Have

• Azure Context: Familiarity with Azure ML deployment and Databricks on Azure (valuable for understanding our broader ecosystem).

• GenAI Deployment: Hands-on experience deploying LLMs, VLMs, or SLMs on GPUs. Familiarity with serving frameworks like Ollama, TGI, or vLLM.Multi-

• Cloud Experience: Experience working across different cloud providers or transitioning workloads between them.

• Tools & Tech: Familiarity with Kafka, Aerospike, Redis, MongoDB, VectorDBs like Qdrant, Pinecone.

Desired Competencies

• Engineering Rigor: Strong coding standards in Python/Scala with a focus on modularity and testing.

• Collaborative Mindset: Ability to translate "Data Science speak" into "Engineering requirements" and vice versa.

• Problem Solving: A knack for debugging complex distributed systems issues (e.g., Spark OOM errors, K8s pod evictions).

• Business Impact: Understanding of how ML models directly affect business metrics and prioritising engineering tasks accordingly.