Senior Software Engineer

Roles and Responsibilities

• Design and build resilient, scalable, online metric measurable ML DL recommender APIs.
• Build cost-effective yet scalable, ultra low latency semi-real-time event streams.
• Build deep learning model pipelines and pipeline abstractions, python libraries partner with applied data scientists to build world class agile, incremental, sequential learners.
• Set up and manage Continuous Integration/Continuous Deployment (CI/CD) pipelines for automated testing, deployment, and model integration
• Experiment and upgrade data science (DS) inference systems, build best in class restful frameworks. Streamline OS and python environment migrations, python version migrations co-own end to end production model service testing, capacity planning, testing, container/service health monitoring and alert automation.
• Co-create DS service asynchronous logging modules, rate limiters & circuit breaker modules, automations for smart service management in kubernetes clusters, and integrate with code/PII data protection modules.
• Build generic and charter centric derived event tables stitching events, features, and user feedback loop with appropriate partitions to stream line sequential or online ML learners, contextual bandit learners etc.
• GPU ops opportunity: Learn and optimise GPU services for encoder/predictor models using Tensor RT, Triton and other frameworks and be a champion of GPU ML inference optimisations across org.
• Opportunity to learn, own and add capabilities to offline ML DL model performance evaluation metric systems (MLflow, Arize, W&B, Neptune etc).
• Build ML/DL observability 2.0 modules with a combination of statistical, causal ML, and LLMs-as-judge.
• Co-own cost optimisation to maximise ratio of impact per user vs infra cost per user.
• Assist data science to improve dev cost efficiencies.
• Opportunity to co-create LLM as a judge framework
• Opportunity to streamline embedding ANNs, HNSW, quantisations and vector databases.
• Opportunity to build services with open sourced SLMs and computer vision models.
• Opportunity to experiment and deploy Agentic AI solutions with pre-trained LLMs and context/prompt engineering principles and best in class RAG/graphRAG/API retrieval mechanisms.
• Build a culture for enforcing strong production coding discipline in APIs, and design and solution documentation.
• Work with data analysts, data experts, product/engg and create custom queries and pipelines such that applied DS can efficiently create A/B metric dashboards and conduct error analyses and faster A/B iterations.
• Strong engineering mindset - build automated monitoring, alerting, self healing capabilities. Review the design and implementation in collaboration with Architects and advocate latest DL/ML engineering practices amongst tech.
• Collaboration: Work closely with the Product Managers, Platforms and Engineering teams to ensure smooth deployment and integration of ML models into Myntra production systems.

Desired skills and experience:

• 3 to 5 Years hands on experience and proven expertise with building end to end, complex & robust complex and large Data Engineering pipelines on PySpark or scala.
• 1+ years experience especially in both ML batch feature pipelines, and also ideally/optionally near-real-time kafka consumer pipelines (Spark Structured streaming or Flink).
• Breadth experience working with data blobs, delta lakes, storage for ML workflows.
• 3+ years experience in ML API and ML pipeline engineering is must
• 2+ years excellent coding experience in Python, pyspark(Python3), Flask/Falcon/FastAPI.
• Solid experience in Kafka consumer/producers, read/write connectors to aerospike/redis DBs.
• Experience with ML orchestration tools (Airflow, Kubeflow, MLFlow)
• Must have experience with Qdrant/MIlvus or other vector DBs.
• Understanding of Architecture and Design of Data Engineering products. Be able to articulate the trade offs .

Nice to have:

• Experience with building feature store pipelines will be a plus
• Experience with CI/CD , API testing and monitoring is a plus
• Solid experience with GPU ops is a plus
• Experience with LLM ops for inference and training is a plus.
• Experience building production grade context /prompt engineering pipelines, production grade multi-agentic AI frameworks is a plus.