AI Engineer

appflame - Ukrainian product-driven tech company building world-class products: Hily, Taimi, AdConnect, Mailkeeper, and more.

About us: • 7 years in the market, 500+ team members, offices in Kyiv, London, Limassol, and a co-working hub in Warsaw. • In 2025, appflame ranked 5th among the top 50 employers in Ukraine according to Forbes and won the “Best Place for Growth” award. • Our apps Hily and Taimi are among the top 5 dating apps in the US with over 70 million users. AdConnect and Mailkeeper focus on building proprietary B2B and B2C solutions.

Our mission: • To put Ukrainian-built products on the global map.

Our goals: • Break into the global top 5 product companies. • Become a unicorn. • Make Ukraine a country where unicorns are born.

We’re looking for bold, driven people who are passionate about building real products and dream of launching and scaling great startups. You bring the ambition — we provide the environment to make it happen.

What you’ll do:

• Design and develop an AI-powered productivity analytics platform — from data pipeline architecture to the final analytical product that helps teams make data-driven decisions. • Build scalable LLM pipelines (Claude, GPT): develop data chunking strategies, implement MapReduce approaches for parallel processing of large datasets, and synthesize results into structured reports and insights. • Create a meta-workflow system where LLMs generate, test, and deploy automation scripts in an isolated environment — with automatic self-correction loops and production deployment without manual intervention. • Develop system-level prompt engineering: build and maintain a library of prompt templates for various analytical scenarios — from summaries and profiles to deep performance analysis. • Build an evaluation framework for AI output quality control: hallucination detection, consistency scoring, regression tests — ensuring the product delivers reliable and reproducible results. • Scale the platform to new domains and analysis types without linear growth in manual effort — through an architecture that allows adding new modules via configuration, not code. • Document AI architecture, define automation specs, and present product insights to stakeholders and clients.

It’s a match if you have:

• 2+ years of experience working with LLMs in production: prompt engineering, pipeline development, API integration — with at least 1 year of hands-on experience with advanced features (tool use, structured outputs, agents). • Strong Python skills (async, dataclasses, type hints, API integrations) and a commitment to writing clean, testable, and maintainable code. • Understanding of MapReduce patterns for LLM processing: ability to choose chunking strategies, organize parallel processing, and aggregate results into a cohesive analytical product. • Experience building agentic systems: tool use, self-correcting loops, multi-agent workflows — and the judgment to know when an agent works better than a rigid pipeline. • Proficiency in SQL and experience working with analytical databases. • English at C1 level — comfortable reading documentation, writing technical specs, and communicating asynchronously. • Ownership mentality: you take tasks end-to-end, debug production issues independently, and iterate to deliver results without micromanagement.

Nice to have:

• Experience with orchestration/automation platforms (Windmill, Dagster, Prefect) — understanding how to build reliable automated workflows. • Knowledge of RAG architectures, vector databases, and embedding pipelines — ability to build systems that work with large volumes of unstructured data. • Experience building evaluation systems for LLMs (LangSmith, PromptFoo, or custom solutions) — understanding how to measure and improve AI product quality. • Familiarity with Databricks / Delta Lake or Snowflake — experience working with modern data platforms. • Experience working at product-driven tech companies or AI startups where you needed to build and iterate quickly. • Understanding of product team metrics (DAU, retention, unit economics) and the ability to connect technical decisions to product impact.

Preferred tech stack:

Core: • Languages: Python (primary), SQL • LLM APIs: Claude API (Anthropic), OpenAI API • Databases: PostgreSQL, ClickHouse • Infrastructure: Docker, Git, FastAPI, Pydantic, pytest, asyncio

Nice to have: • Languages: TypeScript/JavaScript, Bash • LLM APIs: Google Gemini API • Orchestration & Automation: Windmill, Dagster, Prefect, Airflow, Temporal • RAG & Embeddings: LangChain, LlamaIndex, ChromaDB, Pinecone, Weaviate, pgvector, FAISS • Eval & Observability: LangSmith, PromptFoo, Weights & Biases, Arize AI • Databases: Redis, MongoDB • Data Platforms: Databricks, Delta Lake, Snowflake, BigQuery • Infrastructure: Kubernetes, AWS (Lambda, SQS, S3), GCP • CI/CD & DevOps: GitLab CI, GitHub Actions, Terraform

Hiring process: recruiter outreach > interview > test task > final interview > job offer.