Experienced judgment for consequential technical problems
Mikhail Voloshin has been building software for more than 25 years, long enough to watch several technologies arrive as the answer to everything. His career runs through large-scale software engineering at Microsoft and Google, quantitative finance, computational neuroscience, and today’s AI systems. The common thread is how he works. He starts by asking what problem actually needs solving and how anyone will know it’s been solved. Then he picks the simplest tool that can do the job reliably, whether that turns out to be a language model, a statistical method, or a well-written script. Clients get a working system and a clear explanation of why it was built that way.
A career built across domains, not around one tool
The list of employers matters less than what years of complex systems taught him. Every field Mikhail has worked in rewarded the same discipline: define the job precisely, build for the conditions the system will actually face, and finish with something other people can run.
More than 25 years in software engineering.
SOFTWARE ENGINEERING
Microsoft and Google
Building software for very large user bases teaches a lasting respect for reliability. Code that works in a demo and code that holds up under real load are two different products.
QUANTITATIVE SYSTEMS
High-stakes technical decisions
In quantitative finance, a software error is a bad decision with money attached. That work built habits Mikhail still brings to every engagement: test against messy real-world data and know exactly what the system does when it’s wrong.
SCIENTIFIC COMPUTING
Computational neuroscience
Researchers need tools they can explain in a methods section. Mikhail’s scientific work, including the Dendritic Spine Counter presented at the Society for Neuroscience annual meeting in 2021, turned specialized lab needs into purpose-built software that scientists can use and defend in print.
MIGHTY DATA, INC. TODAY
Operational systems and applied AI
Mighty Data, Inc. builds custom systems for small companies with operational problems, from data pipelines to production software. Some include AI. When they do, the language model is called from conventional code with a narrow job and firm limits, an approach Mikhail has released as open-source tooling in the MDI LLM Kit. When conventional engineering is the better answer, that’s what is built.
Judgment is the product of seeing technologies come and go
Every technology boom Mikhail has lived through produced a few lasting systems and a lot of expensive experiments. The difference usually came down to a handful of decisions made early through a thoughtful, experienced lens.
01
Start with the operation
Real work begins with a blocked workflow, too many hours spent on the same project, or a costly handoff. Starting from a preferred technology means answering a question nobody asked.
02
Use the least complicated answer that works
AI may be part of the solution. Conventional software engineering still carries most of the weight: the structure, the reliability, and the guardrails that keep AI inside its assigned job.
03
Build toward a finish line
A useful system is documented, transferable, and owned by the client. If it only runs while its builder stays on retainer, it isn’t finished or sustainable.
Personal judgment, delivered through a company built for the work
When you hire Mighty Data, Inc., you work with Mikhail directly. Our company is the structure that delivers that work and supports it over time.
Mikhail’s role
Mikhail defines the problem with you, makes the key technical calls, and stays accountable for how the finished system is built and whether it holds up.
Mighty Data, Inc.’s role
Mighty Data, Inc. is where systems are scoped, engineered, documented, and delivered. The company stays available for further work whenever that’s useful, though no system it delivers should need it to keep running.
Direct, rigorous, and designed to end in ownership
You work directly with the person making the technical decisions. There is no account manager or middleman relaying messages. Each engagement is designed to settle the uncertain questions early and finish with a system your team can operate.
“Some developers build systems so hard to maintain that the client can never let them go. I think that’s unethical, and it’s not the career I want. My job is to give you honest technical judgment, build something that works, and hand you everything you need to run it without me.”
— Mikhail Voloshin
Direct access to the decision-maker
Clear definitions of success and scope
Engineering and guardrails appropriate to the risk
Source code, documentation, credentials, and ownership
A defined completion point, not permanent dependency
Mikhail enjoys explaining technology nearly as much as he enjoys building it. He has taught natural language processing at Wizard Academy, walked Real Vision viewers through how GPT is built, and put together seminars for companies that want to understand AI before spending money on it. People tend to leave those conversations knowing exactly what they’re buying, which is the point.
Making complex technology easier to understand
Bring Mikhail the uncertainty, not a predetermined technology.
Tell us what you’re trying to improve, where things feel inefficient or uncertain, or where you suspect AI might help. We’ll figure out fast whether Mighty Data, Inc. is the right fit.