I have always been drawn to building things. Not just analysing them — actually constructing something that works, that solves a real problem, that makes the next person's job faster or sharper. That instinct has shaped everything from the automation tools I have built inside global transaction teams to the models I wrote for my Master's thesis to, eventually, this site.
I started in pure advisory — executing buy-side and sell-side engagements across Germany and Europe. Rigorous, formative work. But over time one thing became clear: the most interesting decisions were being made one step above where I was sitting. The advisory layer produces the analysis; the corporate layer uses it to shape strategy. That distinction mattered to me, and I moved deliberately toward it.
Today I lead financial due diligence for cross-border acquisitions — from investment committee presentations to working capital negotiations and SPA outcomes. What I find most interesting is not the process but its edges: where standard frameworks run out of road, where a target's numbers tell one story and its business model tells another. Those are the moments where judgement matters more than methodology.
The writing on this site grows from two things that have shaped how I see the world. The first is technical: spending years thinking about how AI can be applied to financial analysis — my thesis built a working self-learning model for EBITDA adjustment identification — has given me an unusually concrete understanding of what AI actually does and doesn't do, stripped of the hype. The second is personal: I have lived and worked across India, the Netherlands, and Germany. That trajectory gives you a different lens on geopolitics than you get from reading about it. When I write about AI concentration or narrative sovereignty, it is not abstract concern — it is pattern recognition built across cultures, deal rooms, and borders.
This site is where those threads come together. The analysis here is written with the same discipline I bring to a quality of earnings report: source everything, question the framing, follow the incentives, and say clearly what the numbers actually mean.
Designed and built a self-learning Python model capable of automatically identifying non-operating accounts and one-off items from income statements — the central task of any quality of earnings analysis. The model expands its classification logic over time using a growing internal database, becoming more accurate with each engagement.
Built a dedicated financial due diligence automation toolkit integrating key analytical workflows: waterfall chart generation, client-ready workbook formatting, and standardised output templates. Adopted across Accenture's global Transaction Services team.
Developed a VBA-based geolocation tool that identified underperforming US states for digital ad targeting across DACH-region client portfolios. Improved targeting accuracy to 100% and saved thousands of hours of manual review annually.