Machine Learning Engineer

The Synthetic Data Research team is building Ipsos’ next-generation platform for synthetic data and generative AI, turning cutting-edge methods into practical tools that can be used safely and confidently across the business.

We focus on two core products:

• Data Augmentation Workbench: a self-serve internal platform that enables teams to train models and generate synthetic data through secure APIs and streamlined workflows, with evaluation and governance built in from day one.
• Digital Twins: agentic, respondent-grounded LLM “synthetic panellists” designed to simulate behaviours and survey responses, supported by rigorous validation, privacy safeguards, and strong auditability.

Our work sits at the intersection of software engineering, machine learning, privacy, and market research methodology. We collaborate with leading academic institutions (including Stanford University) to ensure our approach is scientifically robust while remaining focused on real-world impact.

Ultimately, our goal is to deliver data collection efficiencies, new product innovation, and defensible scientific frameworks that can scale to thousands of colleagues and clients. We’re a cross-disciplinary group, bringing together market researchers, mathematicians, computer scientists, data scientists, and data engineers, to build capabilities that shape how insights are created in the future.

How you’ll make an impact:

You’ll turn research-grade prototypes into a dependable, governed ML service on GCP. Your work will define how quickly we can iterate on synthetic generation approaches without sacrificing reproducibility, security, or methodological rigor.

In practice, you will:

• Enable Reliable AI: Your pipelines will provide the clean, structured data required to train our core generative models. By ensuring data availability and reliability, you directly support the accuracy and fairness of our machine learning outputs.
• Power LLM Applications: By building robust vector indexing and retrieval (RAG) infrastructure, you will provide our LLM-based synthetic personas with the context and grounding they need to operate effectively and hallucinate less.
• Improve ML Iteration Speed: By optimizing data I/O and standardizing how our models consume datasets, you will eliminate training bottlenecks. This allows our Applied Scientists to run experiments more efficiently and deploy models faster.
• Ensure Data Integrity: In systems relying on statistical weighting and synthetic generation, data quality is critical. Your work implementing strict data contracts and automated validation will prevent bad data from silently corrupting downstream models and evaluation metrics.
• Connect Engineering and Research: You will serve as a key link between traditional data engineering and applied machine learning, translating the complex data requirements of ML research into scalable, maintainable infrastructure.

Tech stack & ecosystem:

You will be the backbone of our ML platform, building everything from user-facing interfaces down to the data layers that feed our models:

• Platform & API: Python, FastAPI, React, TypeScript.
• Data Warehousing & Storage: Google Cloud Platform (GCP), BigQuery, Cloud Storage, Apache Parquet / Arrow.
• Data & ML Orchestration: dbt, Apache Airflow, Kubeflow Pipelines (KFP), Asynchronous Job Queues (Celery/RabbitMQ).
• Unstructured Data & GenAI: Vector Databases (e.g., Pinecone, Weaviate), modern RAG tooling (LangChain, LlamaIndex).
• Data Quality & Contracts: Pydantic, Great Expectations, strict JSON Schema validation.

What You’ll Do:

• Own the API & Platform Layer: Design, build, and maintain the robust backend APIs (FastAPI) that serve as the bridge between our React frontend and our asynchronous, heavy-compute machine learning pipelines.
• Build the User Interface: Develop and maintain features in our React frontend, creating intuitive platform dashboards that allow users to design ML experiments, trigger data augmentation jobs, and visualize synthetic data metrics.
• Architect ML Data Pipelines: Build and maintain high-throughput ETL/ELT pipelines capable of ingesting massive tabular datasets directly into our Kubeflow training and inference workflows.
• Build the RAG Foundation: Develop the data pipelines that power our LLM digital twins. You will handle chunking, embedding generation, and vector indexing of unstructured text to enable highly accurate Retrieval-Augmented Generation (RAG).
• Optimize ML Data I/O: Optimize how our PyTorch models read and write data, leveraging columnar formats (Parquet) and distributed processing to eliminate I/O bottlenecks during training and generation.
• Enforce Strict Data Contracts: Ensure seamless communication between the frontend, backend, and ML workers by implementing strict data contracts (using Pydantic) and automated schema validation.

What you’ll need (role requirements):

Platform & Full-Stack Engineering

• API Design & Backend: Proven experience building robust, highly available RESTful APIs in Python (FastAPI preferred). Experience managing asynchronous workloads and task queues.
• Frontend Development: Solid experience building and maintaining modern, responsive web applications using React and TypeScript.
• Infrastructure & CI/CD: Comfortable working with Git, CI/CD pipelines, Docker, and Infrastructure as Code to deploy platform services reliably.

Data Engineering & MLOps

• Cloud Data Warehouses: Deep expertise in modern cloud data architectures, specifically Google Cloud Platform (BigQuery, GCS).
• Pipeline Orchestration: Hands-on experience with modern data orchestration and transformation tools (e.g., Apache Airflow, dbt) and familiarity with ML orchestrators (Kubeflow, Vertex AI).
• Familiarity with ML Workflows: You understand the data lifecycle of machine learning and know how to prepare data for training, inference, and evaluation.
• Vector Data: Experience or strong familiarity with processing unstructured data and interacting with Vector Databases for semantic search/RAG architectures.

Benefits:

• Flexible & Hybrid working;
• Flexible Benefits platform (e.g.: 7Card, Kindergarten support and many more);
• Additional Vacation days (starting 25 days);
• Referral & Seniority bonus;
• Gifts and events on festive days of the year;
• Bookster;
• Employee Assistance Program;
• Rewarding program;
• Opportunities for professional growth;
• Online Learning Platform;
• Productive & collaborative atmosphere;
• Social and environmental responsibility.

Ready to have an impact? Apply now!