Senior Machine Learning Engineer

📋 Description

• • Own the full machine-learning lifecycle for Portcast’s predictive trade-flow platform, from raw ocean-freight and air-cargo data ingestion to real-time ETA and demand-forecast APIs that power decisions for global logistics giants.
• • Architect, train, and productionize state-of-the-art models for visibility, demand forecasting, freight-audit anomaly detection, and dynamic pricing—handling billions of rows of streaming AIS, port, weather, and commercial data.
• • Design and maintain scalable MLOps pipelines on AWS/GCP using Docker, Kubernetes, Airflow, and CI/CD so every model version is reproducible, monitored, and auto-scaled to sub-second latency for thousands of concurrent users.
• • Implement Retrieval-Augmented Generation (RAG) and AI-Agent workflows that let customers query complex supply-chain scenarios in natural language and receive actionable predictions, turning generative AI into measurable ROI.
• • Lead prompt-engineering best practices, fine-tune open-source LLMs, and build guardrails that ensure accuracy, cost-efficiency, and compliance in a heavily regulated industry.
• • Collaborate daily with product managers, data engineers, and customer-success teams across Singapore, Bangalore, Ho Chi Minh, and Sarajevo to translate C-level pain points into ML features shipped within sprint cycles.
• • Establish rigorous experimentation frameworks—A/B tests, shadow deployments, back-testing against historical voyages—to prove model lift and continuously raise the bar on forecast accuracy (current target: <4% MAPE for ETAs).
• • Drive first-principles thinking sessions, challenging assumptions on feature selection, loss functions, and data labeling to uncover breakthrough insights hidden in messy maritime datasets.
• • Mentor junior ML engineers and data scientists through code reviews, pair programming, and brown-bag sessions, fostering a culture of curiosity and ownership across our remote-first, globally distributed team.
• • Champion data governance and privacy standards, ensuring GDPR, SOC-2, and industry-specific compliance while handling commercially sensitive shipping manifests and customer forecasts.
• • Build real-time anomaly-detection systems that flag fraudulent freight invoices within minutes, saving customers millions in overcharges and cementing Portcast as the go-to audit engine.
• • Create interpretable model dashboards that let non-technical stakeholders understand prediction drivers, increasing trust and accelerating enterprise sales cycles.
• • Stay ahead of the curve—evaluate emerging techniques such as temporal graph neural networks, transformer-based time-series models, and edge-deployment optimizations—to keep Portcast at the cutting edge of predictive logistics.
• • Compress years of learning into months by shipping code that directly moves revenue, customer NPS, and global supply-chain resilience in a lean, high-impact startup environment.