By: Peter P.
Most AI projects fail because nobody in the room can explain what the model is doing, why it produced a different answer on Tuesday than on Monday, or what that variance means for a business that answers to regulators. Hanna Klimushka has spent the last three years turning that gap into a discipline.
A London-based senior Product and Program manager with 14 years in tech, Hanna works where artificial intelligence meets industries that cannot afford to guess: pharmaceutical diagnostics, patient data, government healthcare funding. She is also the founder of PM_ON, a project management education platform, community and agency, through which she has trained more than 500 PMs and delivered work for over 60 clients.
“An AI product isn’t finished when the model works. It’s finished when a clinical director, a compliance officer and an engineer can look at the same output and agree what it means. That translation is the job,” says Hanna.
Starting in the moderation queue
Hanna Klimushka entered technology in 2012 as a moderation team lead at Opinion Corp, a US user-generated content platform operating out of Kharkiv. There was no computer science degree and no accelerated route.
She recalls, “I started in a moderation queue. That teaches you something a degree doesn’t, how a product actually behaves when real people touch it.”
By 2016, Hanna was running Agile delivery at WiserBrand, where she led HIPAA and GDPR compliance delivery alongside US legal counsel, achieved full certification, and strengthened client satisfaction across the portfolio through structured resource planning.
At Ciklum, the global engineering firm, she moved into senior delivery management, leading teams of 25 or more engineers, shortening development cycle times, and sustaining consistently positive client feedback. Proactive RAID-based risk management brought down the number of active project risks and improved cost efficiency.
Scale, then complexity
Relocating to the UK, Hanna Klimushka joined Chaucer, part of BIP Group, as a management consultant, directing a digital transformation program across eight concurrent workstreams in the energy sector with more than 50 full-time contributors. She rebuilt the governance layer with structured reporting, milestone demos and standardized coordination between workstream leads. Senior stakeholder satisfaction improved, missed deadlines fell, and day-to-day operations ran more efficiently.
“Governance is just making assumptions visible while they’re still cheap.”
The translation problem
Since October 2023, Hanna has been Senior Product and Program Manager at Eularis, which builds AI solutions for the pharmaceutical industry. It is here that her particular contribution has taken shape.
She led delivery of an AI-powered pharma diagnostics platform from discovery through scale-up, coordinating between 15 and 40 specialists across engineering, data science, machine learning, QA and DevOps, and working to 8-to-16-week release cycles. More than ten cross-functional discovery workshops with clinical, commercial and technical stakeholders cut down requirement rework and secured full roadmap sign-off in four weeks on a regulated US patient diagnostics platform. Across a SOC 2 certification program, she embedded audit trails and AI/ML regulatory controls.
On a separate engagement, she manages a retrieval-augmented generation system that reads and synthesises specialised medical documentation, currently benchmarked for accuracy on complex clinical content. Much of that work is not technical. It is explaining to pharmaceutical stakeholders why a neural system produces variable output, and building governance that makes the variance manageable rather than alarming.
From this, Hanna developed a set of reusable artifacts now used in client work and taught across her programs: AI explainability report templates, formal AI decision-escalation processes, and a method for translating model metrics such as accuracy, precision, and recall into the language executives budget against.
“A data scientist says accuracy improved one percent. A CFO hears nothing. My job is to say what that one percent means in the language the business already uses. Same fact, two languages.”
Her largest current program is a multi-phase AI platform for a global specialty pharmaceutical company, tracking billions in fragmented US government healthcare funding across all fifty states and converting it, through natural language processing and predictive modeling, into commercial intelligence. Hanna is the named Product Manager, leading data scientists, big data engineers, a solutions architect, UX designers, and business analysts. The discovery phase she managed produced a 130-page solution vision and scope document covering technical architecture, a data readiness scorecard, HIPAA compliance, STRIDE threat modeling and full information architecture.
Building, not only managing
Hanna also ships her own products. From March 2025 to April 2026, she was Senior Product Manager at AI startup Flux AI, taking Craisee, a multi-provider AI creation platform spanning video, image, audio and text, from concept to MVP in under twelve weeks. She defined the credit-based monetization model and tiered pricing, and oversaw a multi-cloud infrastructure decision that brought inference costs down substantially against direct API alternatives. The platform is now live and publicly available.
She then built one herself. Not Now is a prioritization platform putting eight techniques (RICE, WSJF, ICE and MoSCoW among them) into a single interactive toolkit for product and project managers. Its name is her argument in miniature.
“Prioritization is the art of saying ‘not now’. Most teams can’t say it, so they say yes to everything and deliver nothing on time.”
Scaling the practice
PM_ON began as a way to teach what Hanna was learning in live delivery. Its flagship course, Project Management with a Wow Effect, has now completed five cohorts with more than 100 graduates. The community around it runs to almost 400 members, and an agency arm has delivered for over 60 clients. Across LinkedIn, Instagram, Telegram and YouTube, she reaches thousands of practitioners.
She has spoken at UA Online PMDay, a long-running Ukrainian project management conference, at six consecutive editions since 2023, six talks and six expert panels. The subjects track her own work: standing out in the PM job market, consulting briefs, program management at 80-plus FTE scale, AI in everyday PM tooling, whether AI startups have any process at all, and, in 2026, the conference’s state-of-the-field session on where Project and Product management is heading. She has also spoken in Warsaw at the W Tech meetup on hiring managers who fit a company’s DNA.
The organization has begun to carry its own weight. PM_ON was a partner of the Perspektywy Women in Tech Summit in Warsaw, Europe’s largest conference for women in technology, where a member of Hanna’s team presented on the community’s work.
Certified by both Scrum Alliance and Scrum.org at PSM II level, and currently completing PMP, Hanna is deliberate about remaining a practitioner rather than a former one.
“I won’t teach from memory. Everything I teach, I used on a live project that quarter, including the parts that went badly.”
Her position on the industry’s central anxiety is unsentimental: “AI won’t replace Product managers. But a Product manager who works with AI will replace one who doesn’t.”
Learn more at wow-pm-course.com.
