Applied Data Scientist

About AgenticBricks

AgenticBricks is an AI consultancy that embeds expert engineers within enterprise clients to solve complex and important challenges. Their expertise includes agentic systems, intelligent automation, and LLM-driven solutions that deliver measurable business value.

Position Overview

We are seeking a full-time Applied Scientist to join AgenticBricks, supporting a major ecommerce retailer. This role involves full ownership of the machine learning lifecycle, including crafting features, developing and deploying production-grade ML models, and managing scalable inference systems to serve real-time retail operations at high volume. The work is hands-on and focused on delivering production systems rather than exploratory prototypes.

Key Responsibilities

• Develop and maintain features from complex retail datasets, including transactions, product catalogs, user behaviors, supply chain, and operational metrics.
• Create reliable batch and streaming feature pipelines ensuring data correctness, freshness, and reusability across various models.
• Collaborate with feature stores and data infrastructure teams to maintain consistency between training and serving data, troubleshooting train/serve data skew issues.
• Implement, validate, and deploy production models based on live production data rather than limited datasets.
• Establish reproducible training workflows with versioned data and features, automated retraining, and evaluations that determine model promotion.
• Optimize models for scale and cost considerations; design and execute offline and online experiments, including A/B testing, to validate model improvements.
• Build and optimize model serving frameworks for batch, real-time, and low-latency inference under heavy retail traffic.
• Manage inference operations addressing latency, throughput, cost efficiency, autoscaling, monitoring, and drift detection.
• Rapidly diagnose and resolve production issues by connecting operational observations back to feature engineering and training refinements.

Qualifications

• Master's degree or higher in a quantitative discipline (e.g., machine learning, computer science, statistics, applied mathematics) or comparable applied experience.
• Solid foundation in machine learning concepts and statistical skills to rigorously assess model performance.
• Proficiency in Python and standard machine learning/data engineering tools with an ability to write maintainable, production-grade code.
• Proven track record of deploying models fully into production environments at significant scale, covering feature engineering, training, and serving.
• Hands-on experience managing the entire ML pipeline and understanding common failure modes at each stage.
• Strong communication skills capable of explaining methodologies, results, and limitations clearly to non-technical stakeholders.

Preferred Experience

• Background in ecommerce, retail, marketplaces, or large-scale consumer-facing products.
• Experience with feature stores, streaming data pipelines, distributed model training, and model serving systems.
• Knowledge of recommendation engines, search and ranking algorithms, or forecasting techniques.
• Expertise in MLOps practices such as CI/CD for ML, model monitoring, version control, and drift detection in production environments.