AI Model Engineer — Embeddings & High-Performance Neural Architectures
We’re looking for an AI Model Engineer who thrives at the frontier of applied AI, representation learning, and high-performance model design.
We’re seeking an AI Model Engineer to help build the model foundation for AI-native data infrastructure. You’ll work on embedding models, efficient neural architectures, semantic representation, and high-throughput inference systems that make large-scale object storage searchable, understandable, and automation-ready.
This role is not about building a conventional chatbot or application-layer RAG product. It is about designing and optimizing the models that allow data platforms to understand files, metadata, documents, logs, images, and structured content at scale.
You’ll work closely with engineering leadership to prototype, benchmark, and productionize models that can operate over massive volumes of enterprise data with strong performance, practical latency, and real-world retrieval quality.
What You’ll Be Doing
* Design, train, fine-tune, and evaluate embedding models for documents, objects, metadata, and multimodal data.
* Explore high-performance neural architectures, including CNN-inspired models, gated convolutional blocks, efficient sequence models, transformers, and hybrid architectures.
* Build models optimized for semantic search, classification, clustering, tagging, similarity matching, and automated data discovery.
* Optimize inference performance through batching, quantization, distillation, pruning, compilation, and accelerator-aware deployment.
* Evaluate model quality using retrieval, recall, clustering, semantic similarity, and downstream search benchmarks.
* Work with large-scale datasets extracted from object storage, structured files, logs, documents, and enterprise data sources.
* Prototype and benchmark new model architectures rapidly, then help bring the best ideas into production.
* Collaborate with systems engineers to ensure models are deployable at high throughput and aligned with real infrastructure constraints.
* Help shape the AI model layer behind next-generation intelligent storage, search, and data automation.
What We Need to See
* Strong hands‑on experience with PyTorch, TensorFlow, JAX, or similar model development frameworks.
* Practical background in embeddings, representation learning, neural networks, transformers, CNNs, or sequence models.
* Experience evaluating models for retrieval, semantic search, clustering, classification, or similarity matching.
* Strong understanding of model-performance tradeoffs: quality, latency, throughput, memory footprint, and cost.
* Experience with model optimization techniques such as quantization, distillation, pruning, ONNX, TensorRT, or similar tooling.
* Ability to work with large datasets and build repeatable training, evaluation, and benchmarking workflows.
* Strong Python skills and comfort working with production engineering teams.
* Curiosity, technical depth, and the ability to move from research prototype to production‑ready model behavior.
Ways to Stand Out
* Experience building or fine‑tuning embedding models for enterprise search, document AI, code search, multimodal search, or large-scale retrieval.
* Hands‑on work with CNN-based, gated convolutional, or efficient sequence architectures.
* Experience with GPU inference, CUDA, Triton, TensorRT, ONNX Runtime, or accelerator‑aware model serving.
* Background in contrastive learning, Siamese networks, CLIP-style training, sentence‑transformers, or domain‑specific embeddings.
* Experience working with PDF, Office files, logs, source code, Parquet, Avro, image data, or other complex enterprise data formats.
* Published research, open‑source contributions, or strong practical work in retrieval models, model compression, or high‑throughput inference.
* Interest in building models for infrastructure‑scale AI systems rather than only application‑layer products.
Why Join Us?
FastS3 is building AI-native data infrastructure from the ground up. Our vision is to make enterprise data searchable, programmable, and useful to AI systems directly within the data layer.
You’ll join a team working at the intersection of storage, retrieval, machine learning, and high‑performance systems. If you’re excited by the idea of building fast, practical, production‑grade models that help power the next generation of intelligent infrastructure, this is your chance to do it at scale.
We’re growing our team in Madrid and looking for engineers who want to push the frontier of applied AI, not just consume it.
FastS3 is proud to be an inclusive, equal opportunity employer committed to diversity, equity, and accessibility for all.
Application Form
#J-18808-Ljbffr