About Us:
We are building Uzbekistan's leading classifieds platform. Our product helps millions of users buy and sell everything from real estate to electronics. ML is not a supporting tool, but the core of our product: search, ranking, moderation, and recommendations.
The team is actively building the ML infrastructure from scratch: we are selecting technologies, forming processes, and laying the foundation for the next-generation ML platform.
Responsibilities
What you'll be doing:
- Improving search quality: LTR models (LambdaMART/LightGBM), developing and integrating vector search, enhancing the search pipeline (working with suggester, catpred).
- Building ML models for Trust & Safety: fraud detection, ad moderation, NER in messengers, violation scoring.
- Developing the recommendation system - item2item, user2item recommendations.
- Participating in the design and deployment of the ML platform (SageMaker, MLflow, Feast, Evidently, Airflow).
Tech Stack:
Python / CatBoost / XgBoost / Solr / Sentence-Transformers / MLflow / SageMaker / Feast / Airflow / Trino / Iceberg
Requirements
What we're looking for:
- Experience in developing and deploying ML models in production (2+ years): classification, ranking, regression, clustering.
- Proficient in Python and the standard ML stack: scikit-learn, CatBoost/XGBoost, pandas, numpy.
- Understanding of basic statistics and the ability to formulate and test hypotheses correctly.
- Experience working with real-world data: cleaning, feature engineering, handling missing values and class imbalance.
- Ability to work with SQL and large volumes of data (Hive, Trino, Spark or similar).
- Ability to independently take a task from definition to completion – without constant supervision.
- Ability to communicate with the product team: translate business tasks into ML problem statements and explain results to non-technical colleagues.
- AI-first: understanding what a harness is, what skills are, and how to correctly apply AI in development.
Will be a plus:
- Experience in Trust & Safety, content moderation, or fraud detection.
- Familiarity with search and ranking: BM25, LTR, vector search.
- Experience building recommendation systems.
- Experience in Ecom/classifieds.
Why join us:
- Work on a real product with high load and audience in Uzbekistan.
- Opportunity to shape the ML strategy from scratch, not just maintain legacy.
- Small team with high autonomy and a short path from idea to production.
Conditions
What we offer:
- Participation in project development from the start.
- Direct impact on the final product.
- Professional and career growth prospects.
- Development within a team of experienced colleagues.
- Minimum bureaucracy and fast decision-making.
- Individual discussion of work schedule and format with each employee.
