Jakub K. Nowak

0.5 FTE · remote

• Leading a 5-person core team running Poland's AI safety community (~120 monthly active members); responsible for strategy, operations, programme management, and fundraising.
• Organizing events and cross-organizational partnerships in Warsaw and Poznań.
• Conducting ecosystem mapping of the Polish AI safety landscape to identify gaps and coordination opportunities.

Research & Community

• Apart Research Hackathons (2026): (1) AI Manipulation — tested whether LLMs produce more manipulative corporate communications as stakes increase, measuring omission and fabrication rates across 5 models and 3 crisis scenarios. (2) AI Control — fine-tuned DeepSeek-R1-Distill-Qwen-32B to test whether monitor-evasion capabilities propagate through training data (CoT deception transfer).
• SPAR: AI Safety Connect (2024–2025): Consulting on semantic search tooling and product strategy for alignment research datasets.
• ML4Good Bootcamp: Participant (2023), Staff / Teaching Assistant (2024) — led sessions on AI agents and red teaming.
• Talks & presentations: "Intro to AI Safety" at Adam Mickiewicz University (2024, ~40 attendees); internal AI Safety course at DS Stream; multiple AISP community meetup presentations on specification gaming, alignment faking, and AI governance.
• Blog: Epsilon Aligned (Substack, since December 2025) — rationality, EA, and AI safety.

0.5 FTE from April 2025 · remote

• Applied product approach to data tools supporting strategic business decisions, optimizing advertising spend efficiency across supply paths for a Fortune 100 FMCG producer.
• Used A/B testing and statistical analysis of large-volume datasets to drive decisions improving KPIs of programmatic advertising, helping reallocate millions of dollars yearly to higher-quality ad inventory.
• Led development and deployment of ML/LLM solutions using Google Cloud AI Platform (Vertex AI), combining LLMs with traditional ML for production environments.
• Contributed to responsible development of AI-powered products including a RAG-based knowledge assistant and customer service voice bot, handling privacy & safety testing.
• Team leadership: client communication, individual development plans, recruitment, reporting.
• Running LLM & n8n automation training workshops for external audiences (January 2026).

Marketing & User Analytics · remote

• Set up privacy-oriented analytics infrastructure (GTM, Google Analytics, first-party tracker, ETL pipelines, Tableau) and a data warehouse for conversion rate & funnel analytics.
• Led junior analysts and managed analytical product lifecycle. Designed and maintained 20+ dashboards and 20+ ETL pipelines across payment, retail, and marketing data.

Poznań, Poland

• Cross-functional analyst on implementation of a customer-facing Guidewire insurance portal for a UK company, collaborating with UX designers, engineering, and SMEs to deliver accessible experiences for senior (50+) users.
• Helped establish an internal UX Center of Excellence, educating teams on usability and accessibility practices.
• BA, Scrum Master, and proxy Product Owner for an R&D team developing a B2B automated regression testing tool for non-technical users.

M.Sc. in Data Analysis & Processing

Master thesis: Analysis of language changes in news groups and message boards — web scraping, ELK indexing, NLP, time series analysis, data visualization.

M.A. in Finance & Accounting: Risk Management & Insurance

Certificates & Courses

• Product Vision: Certified Product Management Professional — June 2025
• Google: Professional Machine Learning Engineer — April 2025
• ML4Good: ML Engineering & ML Safety Bootcamp — June 2024
• Sages.pl: Text Mining and Natural Language Processing — June 2022