Lead AI Engineer

AT&T

Bratislava, Slovakia

Main requirements:

Deep expertise in full-stack Python, agentic frameworks (e.g., LangGraph, LangSmith), Model Context Protocol (MCP), prompt engineering, context engineering, harness engineering, Retrieval-Augmented Generation (RAG), large language models (LLMs), fine-tuning, and machine learning.

Big plus: Deep knowledge of modern data stack technologies (e.g., Databricks, Snowflake) for pipeline integration and scalable deployment.

As a Lead AI Engineer, you will architect, build, and deploy context-driven, production-ready AI solutions that transform how entire organizations work and deliver value. You’ll leverage the power of LLMs and next-gen AI agents to automate complex workflows, unleashing revolutionary productivity and efficiency. Your work will directly drive the adoption of autonomous AI agents, enabling smarter operations and seamless data pipelines.

Core Responsibilities:

Collaborate cross-functionally with product, data science, and engineering teams to scope, define, and translate complex business needs into technical AI requirements.
Evaluate and select optimal technologies for cloud and on-premise deployments, implementing strategies for scalability, performance monitoring, and cost optimization.
Architect, develop, and integrate scalable agentic AI solutions, including the backend infrastructure required for seamless end-to-end deployment.
Implement testing, monitoring, and evaluation pipelines to continuously improve prompt effectiveness, context handling, and overall agent performance.
Participate in code reviews, contribute to best practices, and mentor peers on AI automation technologies and frameworks.

Technical Skills:

Deep experience with full-stack Python development (FastAPI, Flask; SQL/NoSQL databases).
Deep proficiency with agentic frameworks such as LangGraph and LangSmith, including prompt engineering, context engineering, harness engineering, MCP, RAG, large language models (LLMs), fine-tuning, and machine learning.
Deep experience integrating AI agent solutions into production systems.
Deep experience with cloud infrastructure, containerization (Docker), and CI/CD practices.
Deep experience with AI-powered IDEs and tools such as Visual Studio, GitHub Copilot, Windsurf, Cursor, etc.
Exceptional analytical, problem-solving, and communication skills.

Years of experience:

5+ years of software engineering or data science experience, with at least 2+ years of hands-on experience designing, building, and deploying agentic AI solutions in production.
As a Big Plus – Deep Data Science + LLM/GenAI Data Science Experience:
- Foundations in statistics and machine learning (probability, hypothesis testing, experimental design, bias/variance, regularization) and ability to translate ambiguous problems into measurable ML objectives.
- Experience building and validating predictive models end-to-end (feature engineering, training, cross-validation, model selection, error analysis) using tools like scikit-learn and XGBoost/LightGBM/CatBoost.
- Experience designing rigorous evaluation strategies: offline metrics, online experimentation (A/B tests), guardrails, and monitoring for drift/quality regressions.
- LLM evaluation experience: building automated and human-in-the-loop evals for generation quality (task success, relevance, factuality/faithfulness, toxicity/safety), and using frameworks/patterns like LLM-as-a-judge with calibration and consistency checks.
- RAG-focused data science: measuring retrieval quality (recall@k, MRR, nDCG), diagnosing chunking/embedding tradeoffs, query rewriting, reranking, and grounding/attribution quality.
- Prompt and context optimization using experimental design: prompt A/B testing, prompt regression testing, sensitivity analysis, and systematic prompt/version management.
- Fine-tuning and adaptation experience (as a plus): LoRA/QLoRA, DPO/IPO-style preference optimization, dataset creation/curation, labeling guidelines, and contamination/leakage checks.
- Familiarity with LLM systems performance and cost analytics: token/call optimization, caching strategies, latency/throughput benchmarking, and model routing across providers/sizes.
- Experience with interpretability and debugging for GenAI systems (error taxonomy, hallucination analysis, attribution, and root-cause analysis across retrieval/model/prompt layers).
Additional Plus – Deep Experience with Data Engineering Stack (Snowflake, Databricks).