Machine Learning Engineer | Q-Commerce

About the role

We build the intelligence layer behind a fast-moving q-commerce business — search and discovery, recommendations, delivery time prediction, demand signals, and LLM-powered assistants for customers and support agents. You'll own problems end to end: framing the business question, building the model, shipping it to production, and proving it moved a metric.

What you'll work on

• Build and ship ML models on real Persian-language, high-volume transactional data: ranking and semantic search, recommendation, forecasting, ETA prediction, user segmentation and profiling.
• Take models to production — feature pipelines, training/inference services, monitoring, retraining. Ownership doesn't end at the notebook.
• Build LLM-based systems: retrieval-augmented pipelines, agentic workflows, embedding and vector search, evaluation harnesses to prove they actually work.
• Design offline and online evaluation — A/B tests, ranking metrics (nDCG, MRR, [removed phrase]), business KPIs — and be honest about what the numbers say.
• Partner with product, commercial, and operations teams to translate ambiguous goals into modeling problems.

What we're looking for

• 3+ years building ML or data science systems that reached real users.
• Strong Python and SQL. You're comfortable in a large relational warehouse and can write queries that don't melt it.
• Solid grounding in ML fundamentals — you can explain why a model works, not just call .fit().
• Experience deploying something: APIs, containers, batch pipelines, orchestration. You know the difference between a model and a service.
• Clear communication. You can defend a design decision to an engineer and explain the tradeoff to a business stakeholder.

Strong pluses (any of these)

• Search, ranking, or recommender systems at scale.
• LLM application work: RAG, agent frameworks, prompt and eval engineering, fine-tuning.
• Vector databases, embedding model evaluation, ASR/speech.
• ClickHouse, Spark, Airflow, or similar data infrastructure.

How to think about fit

We're hiring across a spectrum. Some of you lean toward analysis and modeling (framing problems, experimentation, statistical rigor). Some lean toward ML engineering (pipelines, serving, latency, reliability). Some lean toward AI/LLM systems (retrieval, agents, evaluation). We're not looking for someone who is all three at once — tell us where your center of gravity is and where you want to grow.