Senior AI ML Engineer Python Scientific Computing

📋 Description

• • You will architect and scale the core machine-learning platform that powers Swayable’s mission to measure how effectively media changes minds. Every day, millions of data points flow through systems you design, turning raw opinion shifts into actionable insights for political campaigns, global NGOs, and Fortune 100 brands.
• • Own end-to-end model lifecycle—from research-grade prototypes in Jupyter to production-grade services running on Kubernetes. You’ll profile CPU-bound scientific code, rewrite hot paths in Cython or Numba, and push vectorized Numpy/Scipy routines to their limits so that experiments finish in minutes, not hours.
• • Lead the build-out of a distributed feature-engineering layer that ingests survey, demographic, and behavioral data, then transforms it into training-ready matrices with Pandas and Dask. You’ll design idempotent pipelines orchestrated by Celery workers, enforce data-quality contracts, and implement real-time back-pressure so that spikes in traffic never stall the research team.
• • Design and tune ensemble models (gradient boosting, probabilistic neural nets, hierarchical Bayesian samplers) that predict persuasion outcomes with calibrated uncertainty. You’ll run large-scale cross-validation studies, leverage scikit-learn’s ecosystem, and invent custom metrics that quantify causal impact rather than mere correlation.
• • Create a high-throughput Flask micro-service layer that exposes model predictions and counterfactual simulations via GraphQL. You’ll write OpenAPI specs, add observability with Prometheus/Grafana, and maintain <50 ms p99 latency while serving thousands of concurrent API requests.
• • Collaborate with social-science PhDs to translate experimental designs into runnable code. You’ll pair-program to encode complex survey logic, automate power-analysis calculations, and ensure every statistical assumption is testable and reproducible.
• • Build a real-time dashboard in Vue.js + Node.js that surfaces live experiment results to non-technical stakeholders. You’ll push aggregated metrics into MongoDB, craft reactive visualizations with D3, and enable one-click export to Excel and Google Slides for rapid decision-making.
• • Establish MLOps best practices: containerize training jobs, version datasets with DVC, store model artifacts in S3, and trigger automated retraining when drift is detected. You’ll champion CI/CD pipelines that let any data scientist deploy safely on day one.
• • Mentor junior engineers through code reviews, architecture sessions, and weekly “lunch-and-learn” talks. You’ll create internal documentation that turns tribal knowledge into living wikis, and you’ll foster a culture where scientific rigor meets engineering pragmatism.
• • Partner with product managers to roadmap next-generation capabilities—think synthetic-control experiments, uplift modeling, and privacy-preserving federated learning. You’ll estimate effort, prototype MVPs, and pitch technical trade-offs to executives in plain English.
• • Contribute back to open-source: upstream performance patches to scikit-learn, release internal utilities as PyPI packages, and speak at PyData or SciPy conferences to share how Swayable marries social science with scalable Python.
• • Ensure ethical AI by embedding fairness audits, bias-detection tests, and transparent model cards into every release. You’ll work with legal and policy teams to meet emerging regulatory standards while still delivering rapid insights to clients on tight deadlines.