I asked Gemini who I am and why someone should hire me for ML. It gave me a highly accurate, yet 10-year-old version of myself.
I was recently approached by an ML company setting up shop in Zurich. To be honest, I’m still not sure if it was a real person, an agent, or just a sophisticated scouting bot. But it sparked the question: What exactly do they see when they look at me?
I know I can provide crazy value, but within a certain scope. I also know my digital footprint is sparse. To find out what the world thinks of 'James Litsios,' I went into an anonymous Gemini session and asked a simple question: 'Why would someone hire this man for an ML project?'
The answer I got shocked me: a highly accurate "slice" of a man who existed ten years ago.
Let's start with Gemini's response (with some annotations in Blue).
Then I will update you on how I have evolved my core beliefs from the Gemini expressed basis.
Gemini Pro's reply to: Who is James Litsios and why would someone hire him to develop a ML application?
Dr. James Litsios is a veteran technologist, architect, and leader based in Zurich, known for his expertise in high-performance computing, financial engineering, and distributed systems. He is perhaps most recognized as a co-founder of Elevence Digital Finance and a key architect behind ..ML (.. Modeling Language) (my NDA doesn't allow me to repeat these claims, so I removed two letters and two words here).
If you are considering hiring him to develop a Machine Learning (ML) application, his value lies less in "basic model training" and more in the architectural integrity, mathematical rigor, and scalability of the system.
Why Hire James Litsios for ML Development?
- High-Performance Engineering & Scalability
- Litsios has a deep background in quantitative finance and automated trading (notably as the long-time CTO of Actant). ML applications often fail because they can’t handle real-time data or scale efficiently. His experience with multi-core reactive systems, non-blocking queues, and GPU support ensures that an ML model isn’t just a "notebook project" but a production-ready engine capable of handling massive throughput.
- Mathematical & Algorithmic Rigor
- His academic and professional history is rooted in complex simulations and solvers.
- Scientific Computing: He holds a PhD from ETH Zurich with a focus on modeling and simulation.
- Advanced Math: His work has involved automatic differentiation and multi-grid interpolations—concepts that are foundational to how modern neural networks are optimized and trained.
- Privacy & Distributed Execution
- In the modern ML landscape, data privacy is a massive hurdle. Litsios holds patents in privacy-preserving shared execution. If your ML application requires:
- Federated learning (training on data you don't "see").
- Secure multi-party computation.
- Blockchain-integrated AI logic.
...his specific expertise in "smart" distributed contracts and privacy-preserving protocols is nearly unparalleled. - Language & Domain-Specific Logic
- Litsios is a "language person." He helped create ..ML (again two letters removed) to describe complex rights and obligations. In ML, this translates to System Design: building the domain-specific languages or frameworks that allow data scientists to express complex business logic without breaking the underlying infrastructure.
Summary Table: The "Litsios" Advantage
| Category | Skillset |
| Infrastructure | Building the high-speed "pipes" and GPU-accelerated environments ML requires. |
| Integrity | Applying "Smart Contract" levels of precision to ML workflows to ensure auditability. |
| Security | Implementing privacy-preserving layers so ML can use sensitive data legally. |
| Leadership | Bridging the gap between raw research (PhDs) and commercial product delivery. |
Bottom Line: You wouldn't hire James Litsios just to "clean a dataset." You would hire him to architect the entire ecosystem—ensuring your ML application is mathematically sound, cryptographically secure, and fast enough to beat the market.
Me reviewing Gemini reviewing me
Gemini actually got right much of how I engineer systems. For the record, I'll complement Gemini's view with the follows:
- High-Performance Engineering & Scalability: Closing the Loop
- Yes, spot on, Gemini noted my background in multi-core reactive systems and non-blocking queues, providing tools to some of the major derivative market making firms.
- Today, the industry moves toward hardware systems with perfectly synchronous execution model to bypass that lock based bottlenecks. The consequence is that the same kind of tiering semantics found in early distributed contract models, now help synchronize and orchestrate what was previously a lock contention execution bottlenecks. It closes the loop, as the concept of semantic orchestration has roots in my challenges in modeling asynchronous trading streams.
- Mathematical & Algorithmic Rigor: The Era of "Lazy Rigor"
- Yes, however rigor only when needed! And in fact I have adopted "lazy rigor" as best approach.
- I prefer "loose" math written in a DSL that gets formally checked by automated tools. As an engineer, math remains the most powerful tool I have.
- Privacy & Distributed Execution: The Shift to Deep Semantics
- "nearly unparalleled expertise in 'smart' distributed contracts and privacy-preserving protocols"... from 2015 to 2019! (see bearingpoint-spin-off-elevence)
- My current focus is less on privacy and more on deepening semantics. One way to look at this is terms of how many phases towards finalization do you embed in your core semantics? Two, as in a classical database? 3, 4, 7, 8, even more? It sounds like overkill, yet building those deeper phases is the only way to give ML-driven logic absolute certainty.
- Language & Domain-Specific Logic: Bridging the Gap
- Litsios is a "language person." Wow, I agree! Even math is more the "how of semantics" to me than about truly caring about the math as classical mathematician.
- I spent decades obsessing over how to view software as math, from using codified functional programming to using probabilistic formal methods, because I have experienced enough "miserable quality" of standard software development to know that they will not meet the demands of the future.
Bottom Line 2026
Let's sum it up:
- Model Driven High-Performance Engineering
- Today, true architectures are models in high-dimensional, multi-tiered, semantically rich streaming spaces. We are past the "simple pipe" concept. (See Why Use Arrows and PowerPoint to Express Software Designs? and Math over Types)
- Lazy Rigor for Math Everywhere
- I believe "lazy rigor" is the future. We don't need more manual proof; we need DSLs that handle the heavy lifting of formal checking. Math is the ultimate tool, yet the framework does the enforcing.
- Deep Semantics
- The true advantage is in the ongoing multi-phase semantics of finalization. It is not the past, nor the hypothetical future. When AI executes logic, standard database finality isn't enough.
- Unburdened Software Evolution
- We write our new future from scratch, we are no longer held back by past implementation choices.
To finish off, I asked Gemini to illustrate its statement about me!
All original content copyright James Litsios, 2026.