About
Kacper Wikieł · AI/ML Systems Architect · WarsawI've been writing Python since 2014 and building ML systems professionally for the last three years. Before going independent, I worked at PwC Digital & AI, led a PLN 9.3M NCBiR-funded R&D project, and built GenAI infrastructure used by P&G globally. Also: Lecturer at University of Warsaw (AI Applications in Science). International Young Physicists' Tournament participant. Physics background before engineering.
What I do now: end-to-end ML projects for companies that need systems in production, not proofs of concept. I specialize in compliance ML (adverse media, AML), industrial inspection (NDT, signal processing), and enterprise GenAI (RAG, agentic systems).
I'm comfortable presenting to CFOs and debugging CUDA kernels in the same week. I prefer to understand problems deeply before proposing solutions, and I care about shipping things that work—not impressive demos that fail in deployment. Most "AI projects" fail not because of model performance but because of integration, data quality, or misaligned expectations. I've seen this pattern at Big4 scale and startup scale.
What makes me different: I reverse-engineer proprietary formats others treat as black boxes. I bring Big4 methodology without Big4 overhead. I take full ownership—no handoffs, no "that's someone else's problem."
Projects I take: Clear metrics (F1, mAP, precision/recall). Direct access to decision-makers. Real problems where ML is actually the right solution. No AI theater.
Selected Work
End-to-end adverse media screening system for AML compliance. NER + classification pipeline for entity-risk detection across news sources.
NER Classification AMLEnterprise RAG system for internal knowledge base. LangGraph orchestration with custom chunking strategy that significantly improved retrieval relevance over baseline.
RAG LangGraph LLMsSignal processing and ML pipeline for ultrasonic non-destructive testing. Defect classification, automated reporting, integration with engineering workflows.
Signal Processing CNN CVGenAI strategy consulting for food-tech startup. Technology assessment, use case identification, implementation roadmap.
GenAI StrategyTech lead on government-funded R&D project for Open Banking fraud prevention. Substrate-based chain with tamper-evident audit logs, KYC/AML rule engine. Led 4-person dev team. Delivered working prototype to banking partner.
Blockchain Compliance RustExperience
End-to-end project ownership: scoping calls with executives, production code, deployment, and ongoing maintenance.
Reverse-engineered proprietary ultrasonic data formats. Built visualization and analysis tooling. Feature extraction pipeline for defect detection.
Built core backend for P&G's internal GenAI platform used globally. Full RAG pipeline: ingestion, embeddings, retrieval, response generation. Designed parallelization framework for batch inference at enterprise scale.
Earlier experience
PLN 9.3M NCBiR-funded project. 4-person dev team. PSD2, transaction monitoring, regulatory reporting.
Corporate Python courses (40+ hours).
ETL pipeline for Ethereum analytics. ERC-20 bytecode parsing, token discovery engine.
Skills
Certifications
- Google Associate Cloud Engineer (2023–2026)
- Google Cloud Digital Leader (2023–2026)
- Microsoft Certified: Azure AI Engineer
Education
- Polish-Japanese Academy of IT – B.Eng. Information Technology (2026)
- University of Warsaw – Physics (2012–2014)
Other
- Lecturer – AI Applications in Science, University of Warsaw
- International Young Physicists' Tournament participant, 2012 (source)
Contact
For project inquiries: k.wikiel@gmail.com
LinkedIn ·
GitHub ·
CV (PDF)
Best fit: companies with clear success metrics, direct decision-maker access, and real problems where ML is the right solution.