Google ile giriş. Kart gerekmez.
About TEGNA
TEGNA Inc. helps people thrive in their local communities by providing the trusted local news and services that matter most. With 64 television stations in 51 U.S. markets, TEGNA reaches more than 100 million people monthly across the web, mobile apps, streaming, and linear television. Together, we are building a sustainable future for local news.
We are looking for a Data Scientist to help build the data science and applied AI capabilities behind TEGNA’s content platform, the systems that turn a high volume of news and media content into insight, enrichment, and better experiences for millions of users. You’ll work with large, messy, text-heavy datasets and use NLP and large language models to turn them into production solutions that make content more discoverable, better organized, and more useful to editorial, product, and audience teams. You’ll be a technical anchor on the team: framing ambiguous business problems as data science problems, setting the bar for rigor in experimentation and evaluation, and mentoring analysts and mid-level data scientists. You’ll partner with engineering and cloud teams on how models and LLM-based services are deployed, monitored, and operated in production.
Responsibilities:
- Design, build, and evaluate data science solutions that apply NLP and LLMs to content problems such as classification, tagging and taxonomy, entity extraction, summarization, similarity and recommendation, and semantic search.
- Develop LLM-powered solutions, including prompt design, retrieval-augmented generation (RAG), embeddings and vector search, and fine-tuning where it is warranted, with a pragmatic eye on cost, latency, and quality.
- Own evaluation. Define metrics, build benchmark and golden datasets, and run offline and online experiments (including A/B tests) to show that a solution works, and keep checking that it keeps working.
- Frame ambiguous business and editorial problems as well-defined data science problems, and help convert epics and tech briefs into actionable, well-scoped stories alongside the team lead.
- Work with engineers to take models and prototypes into production as reliable services, covering monitoring, drift detection, guardrails, and responsible-AI practices such as hallucination, bias, and content-safety controls.
- Explore, clean, and model large structured and unstructured datasets, and communicate findings and trade-offs clearly to both technical and non-technical stakeholders.
- Set standards for reproducibility, experiment tracking, code quality, and documentation across the team.
- Mentor analysts and mid-level data scientists, raising the overall technical maturity of the team.
- Partner with engineering and cloud teams (who own infrastructure realization, CI/CD, and standards).
- Work in an Agile/Scrum team with disciplined source control.
Requirements:
- 5+ years of professional experience in data science or applied machine learning, with models or analytical solutions delivered into production.
- Strong proficiency in Python and the core data science stack (pandas, scikit-learn, and a deep learning framework such as PyTorch), plus solid SQL.
- Strong grounding in statistics, experimental design, and model evaluation, and the judgment to choose the simplest approach that solves the problem.
- Hands-on experience with NLP (text classification, named entity recognition, topic modeling, embeddings, semantic similarity) and with modern transformer-based approaches.
- Practical experience building with LLMs: prompt engineering, RAG, embeddings and vector databases, and evaluating LLM output quality. Experience with Hugging Face or comparable platforms is expected.
- Experience taking data science work to production in a cloud environment (AWS, GCP, or Azure), and an understanding of MLOps fundamentals: versioning, monitoring, reproducibility, and retraining.
- Ability to communicate clearly with stakeholders and explain model behavior, limit