Senior ML / Search & LLM Ops (Contractor)

About the role

RavenPack, a big data analytics provider for financial services with 23 years of experience, is launching a new suite of GenAI and SaaS services for financial professionals. It is looking for a contractor to join its internal Search & Recommendation engineering team, working directly on the search and LLM stack. The initial engagement is a 3 to 6-month contract with possible extension, and the working model is that of an independent contributor.

Key responsibilities

• Implement and evaluate targeted open-source LLM adaptations (PEFT, LoRA, QLoRA, distillation) for financial contexts, and prototype preference alignment mechanisms (DPO/PPO)
• Benchmark and optimize model serving for low latency and high throughput, applying quantization (AWQ, GPTQ) and using Triton, TEI and vLLM
• Prototype advanced search techniques such as Matryoshka embeddings, late interaction models and hybrid search pipelines
• Pack training, evaluation and serving scripts into reproducible deliverables using AWS SageMaker and Docker
• Set up synthetic data generation and automated evaluation harnesses (Opik, LLM-as-a-judge) to measure cost, latency and quality trade-offs

Required profile

• Master's or PhD in Computer Science, Machine Learning or a quantitative field, or equivalent practical experience
• Proven track record of production-grade ML models and PoCs in Search, IR or LLM infrastructure
• Fluent written and spoken English; European legal working status and EU timezone alignment required
• Ability to work autonomously, deliver well-documented modular code and validate ideas quickly through empirical testing

Required skills

• Python, PyTorch and the Hugging Face ecosystem (Transformers, PEFT, Accelerate)
• Model quantization, VRAM optimization and high-performance serving frameworks (Triton, TEI, vLLM)
• AWS SageMaker, MLflow, Opik
• Bonus: direct RLHF or DPO experience, and familiarity with financial market data

Compensation is a competitive daily or project-based contract rate commensurate with experience.