Experience: 3+ years of ML research experience (academia or industry).

About Solenya

Solenya is building the AI infrastructure layer for modern e-commerce. We help merchants better match products to customers so shoppers can discover the right products faster. This drives higher revenue, stronger engagement and better customer lifetime value.

We’re an early-stage team working closely with leading retailers and marketplaces to shape the future of AI-driven commerce.

The Role

Our platform primarily runs on a core pre-trained model that combines images, text, users and products, so a merchant gets search and recommendations working on day one, with no training on their data. You'll decide how that model gets better: what we train, how we train it and how we know it worked.

You'll work directly with the leadership team and our ML engineers, and your models go into production, serving millions of requests a day.

What You'll Do

Fine-tune multimodal embedding models for search and recommendation
Explore parameter-efficient and RL-based training, such as prompt tuning, LoRA and RL-based training with LLM-judge rewards
Research new discovery problems: composed image retrieval, personalised search, zero-shot sequential recommendation
Build evaluations we can trust, including checking LLM judges against human labels and benchmarking on public datasets
Run clean, reproducible experiments and turn the winners into code our engineers can ship

What We're Looking For

Strong deep learning fundamentals and hands-on PyTorch experience
Research experience in representation learning, information retrieval, recommender systems, CV or NLP
Sceptical of your own results: you care more about whether the eval is right than about the number
Comfortable writing production-quality Python, not just notebooks
High EQ, low ego, and happy in an early-stage environment

Nice to have: publications, experience with RL fine-tuning or preference optimisation, multimodal or vision-language models, or bandits.

Our Stack

PyTorch, Hugging Face (Transformers, PEFT, Datasets), Ray, MLflow, DVC, BigQuery, Polars and GPUs on GKE.

What Success Looks Like

In your first 3–6 months, you'll:

Take at least one new research direction from idea to a working experiment
Make our evaluation suite something the whole team trusts
Contribute to a new model that beats one of our previous models on internal evals

Why Join Solenya

Your research goes into production, not just into a paper
Work directly with the leadership team
High ownership over the direction of our core models
Fast learning and a steep growth curve

Interested? Email us at careers@solenya.ai.