Senior ML, Search & LLM Ops

This is a fully remote position, open to applicants in Spain.

📋 Description

• Perform hands-on development, rapid prototyping, and plug-and-play enhancements for search and LLM systems.

• Develop and assess domain-specific adaptations of open-source LLMs using PEFT, including LoRA and QLoRA, along with distillation tailored for financial applications.

• Create prototypes for preference alignment techniques such as DPO and PPO.

• Benchmark and enhance model serving to achieve low latency and high throughput.

• Utilize AWQ and GPTQ quantization methods while employing Triton Inference Server, TEI, and vLLM.

• Prototype and test advanced search methodologies, including Matryoshka embeddings, late interaction models, and hybrid search pipelines.

• Integrate structured and unstructured data within hybrid search pipelines.

• Organize training, evaluation, and serving scripts into clean, reproducible outputs using AWS SageMaker and Docker.

• Establish synthetic data generation and automated evaluation frameworks such as Opik and LLM-as-a-judge.

• Evaluate cost, latency, and quality trade-offs for delivered proofs of concept.

• Contribute independently while collaborating with the internal Search & Recommendation engineering team.

⛳️ Requirements

• Master’s or PhD in Computer Science, Machine Learning, or a related quantitative discipline (or equivalent practical experience).

• Demonstrated success in delivering production-level ML models and proofs of concept in Search, Information Retrieval (IR), or LLM infrastructure.

• Extensive hands-on experience with Python, PyTorch, and the Hugging Face ecosystem (Transformers, PEFT, Accelerate).

• Practical experience in model quantization, VRAM optimization, and high-performance serving frameworks (Triton, TEI, vLLM).

• Familiarity with AWS SageMaker for training and deployment.

• Experience with experiment tracking and monitoring tools (MLflow, Opik).

• Capability to work independently and produce well-documented, modular code.

• Skill in rapidly validating concepts through empirical testing.

• Proficient in English communication skills, both written and verbal.

• European legal working status is mandatory.

• Alignment with EU time zones is necessary.

• Direct experience with RLHF or Direct Preference Optimization (DPO) is an advantage.

• Familiarity with financial market data and processing of financial text is a plus.

🏝️ Benefits

• Opportunity for contract extension based on project milestones and outcomes.

• Competitive daily or project-based contract rate based on experience.

• Fully remote working environment.

• Equal opportunity workplace that values diversity.