ml engineer for app review intelligence

AppFollow is an app review management and ASO platform that helps app developers, product managers, marketing teams, and customer support teams manage app and game data, improve ratings and app store rankings, and increase user loyalty. Its AI capabilities automate routine work with user feedback through categorization, summarization, reply generation, semantic search, anomaly detection, and conversational insights.

• Own AI/ML features end-to-end, from research and prototyping through production, monitoring, and iteration;
• Design and build LLM-powered features using reviews and ratings data, including feedback categorization, summarization, reply generation, semantic search, anomaly detection, and conversational and agentic scenarios;
• Work with commercial LLM APIs and open-source models, including model selection, adaptation, fine-tuning, and deployment;
• Build and maintain pipelines for model training, fine-tuning, and quality evaluation, including datasets, metrics, offline evaluations, LLM-as-a-judge, and A/B tests;
• Develop RAG and semantic search capabilities, including embeddings, vector storage, and retrieval quality;
• Optimize LLM inference quality, latency, and cost in production;
• Track state-of-the-art developments in NLP and LLMs, run experiments and POCs, and turn promising results into product features;
• Collaborate with backend, product, and platform teams;
• Contribute to the overall system architecture;
• Write efficient, testable, secure, and documented code.

• Have 5+ years of software development experience;
• Have strong production Python and asyncio experience;
• Have 3+ years of hands-on ML/NLP experience with models shipped to production;
• Have practical experience with LLMs, including prompt engineering, RAG, fine-tuning open-source models with LoRA/PEFT, and working with commercial APIs and self-hosted models;
• Have experience building model quality evaluation processes with metrics, evaluation datasets and pipelines, and A/B testing;
• Be confident with the PyTorch and Hugging Face ecosystem, including transformers, datasets, and PEFT;
• Be proficient in FastAPI for API development;
• Have strong SQL skills with MySQL or PostgreSQL and experience with ORM frameworks, preferably SQLAlchemy;
• Have experience with unit testing using pytest;
• Have Upper-intermediate English or higher;
• Nice to have: serving open-source LLMs in production with vLLM, TGI, or Triton, GPU infrastructure, vector databases such as pgvector, agentic and orchestration frameworks such as LangChain and LangGraph, evaluation and observability tooling such as MLflow and Langfuse, data processing pipelines and automation such as Airflow and Prefect, cloud-based services such as AWS, NoSQL databases such as MongoDB, message brokers such as RabbitMQ and Kafka, classical ML/NLP including text classification, clustering, and topic modeling, open-source contributions, publications, or pet ML projects.

• EU only;
• Paid vacation and sick leave;
• Days off for special occasions by prior agreement;
• Health insurance;
• Equipment reimbursement;
• Home office moderation bonus;
• Stock options bonus according to the employee stock ownership plan;
• Monthly All Hands in company hubs in Helsinki, Belgrade, Tbilisi, Batumi, and Yerevan;
• Executive-level visibility into how the company is run and performing;
• Dedicated support and fast-track onboarding.