Oily yeast is the sort of phrase that sounds more at home in a neglected refrigerator than a patent office. To Max Mikheev, it is an industrial proposition. The yeast in question, Yarrowia lipolytica, is naturally good at accumulating oils. Mikheev’s company, BIOMEDICAN, has spent years engineering it to produce compounds that are scarce, expensive, or awkward to extract from plants. His explanation is almost disarmingly plain: “In nature, cannabis plants produce cannabinoids in drops of oil, so it’s exactly what we’re trying to produce inside our yeast.”
Plain explanations are useful in synthetic biology because the work beneath them is ferociously specific. Genes must be chosen, inserted and balanced. Enzymes must hand molecules from one step to the next without creating too many unwanted byproducts. A flask that behaves beautifully can become temperamental in a larger vessel. After fermentation comes separation, purification and the unromantic discipline of making the same material again.
Mikheev has been preparing for complicated systems since long before BIOMEDICAN was incorporated in California in 2017. His public record begins in the scientific institutions of Novosibirsk, where he trained in molecular biology. He worked on virology and genomics, co-authored papers on viral identification, and learned to think at the level of sequences. Later research posts in Pittsburgh and Bethesda moved him into computational and experimental work. The instruments changed. The fascination with systems did not.
The useful detour through software
One revealing stop was SPARK, short for Simple Platform for Agent-based Representation of Knowledge. Mikheev was among the researchers credited on work describing the platform in 2010. Agent-based modeling treats a complex system as a population of individual actors following rules. You specify the parts and their behavior, then watch larger patterns emerge. It is a computer scientist’s way of respecting biological messiness.
Then came BioDatomics. Mikheev co-founded the company and served as chief technology officer, bringing large-scale data methods to genomic analysis. Public descriptions called it the first Hadoop-based big-data genomics platform. Whatever priority one assigns to that phrase, the venture placed him at a useful intersection: scientists drowning in sequence data, software capable of distributing the load, and a market beginning to understand that biology had become an information problem.
That period makes BIOMEDICAN look less like a left turn. Wet biology taught Mikheev what the pieces were. Modeling taught him to think about their interactions. Data platforms taught him to turn expert workflows into systems other people could use. A fermentation company asks for all three instincts, then adds stainless steel and a balance sheet.
“The yeast which we are using is for the production of oils and the production of cannabinoids.”Max Mikheev
Choose a host that likes the assignment
The shrewdest part of BIOMEDICAN’s thesis may be its choice of organism. Yarrowia lipolytica is an oleaginous yeast - “oil-forming,” without the Latin flourish. Instead of forcing a generic laboratory workhorse to become something entirely alien, the company starts with a microbe whose metabolism already leans toward lipids. Cannabinoids are oil-soluble. The fit does not solve the engineering, but it gives the engineering a favorable slope.
Mikheev’s patents with scientist Difeng Gao describe yeast containing genes for several linked tasks: making a precursor called GPP, producing olivetolic acid, building a supply of Hexanoyl-CoA, and completing steps toward a cannabinoid or its precursor. Read as a list, it is alphabet soup. Read as a factory diagram, it is a sequence of workstations. Each must receive material, do its job and pass the result along.
The first U.S. grant, number 11,149,291, arrived in October 2021. A related grant, number 11,939,613, followed in March 2024. A continuation application was published that August. Patents are milestones, not factory inspection reports. They say that an invention has crossed a legal threshold of novelty and description. They do not certify yield, price, market demand, or commercial scale. Still, they are unusually concrete breadcrumbs in a sector fond of misty promises.
A platform must earn the noun
BIOMEDICAN has presented its technology as more than a cannabinoid process. It has also described work on astaxanthin, a red pigment used in aquaculture and other products. In the company’s account, researchers tried a series of genetic designs, then spotted crystals in one batch. Analysis showed a mixture dominated by astaxanthin, alongside beta-carotene and canthaxanthin. It is the kind of laboratory moment that tempts a tidy origin myth: after months of invisible molecular labor, the result announces itself in color.
The broader ambition matters. A single engineered strain can make a product. A platform should make the next product easier. The reusable assets are not merely genes. They include the host organism, transformation methods, pathway knowledge, analytical routines, fermentation recipes, purification tricks and the institutional memory of what failed last Tuesday. In biotechnology, the word “platform” is often handed out before it has done the work. BIOMEDICAN’s astaxanthin project was an attempt to show that its toolkit could travel.
Mikheev’s career makes him temperamentally suited to the platform argument. BioDatomics was software for repeatable analysis rather than a single study. SPARK was an environment for building models rather than one model. BIOMEDICAN proposes a biological chassis rather than one molecule. Again and again, he moves one level above the immediate task and tries to build the machinery that can repeat it.
The gap between diagram and drum
There is a reason the synthetic-biology story is so seductive. On a slide, the agricultural supply chain folds into a loop: sugar enters, yeast grows, product leaves. Weather disappears. Acreage disappears. The long plant cycle shrinks to fermentation time. A stable process can run in many places and on many schedules. The sketch feels less like farming and more like programming.
Reality insists on footnotes. Cells respond to crowding and heat. Oxygen moves differently through a large tank than a small flask. A pathway that produces a molecule may also burden the organism producing it. Downstream purification can swallow whatever savings fermentation created upstream. Customers care about specifications and dependable lots, not the elegance of the genetic cassette. Scale is where biology submits its invoice.
Mikheev and BIOMEDICAN have publicly forecast lower costs, high purity and reduced resource use. Those are the company’s targets. The patent record shows the engineering route; it cannot settle the economics. That distinction makes the story more interesting, not less. The company is not merely trying to prove that yeast can make a molecule. It is trying to prove that a biological method can become a competitive manufacturing system.
This is where Mikheev’s software years offer a quiet advantage. Software founders learn to separate a demonstration from a product: the demo proves a function, while the product survives strangers, repetition and inconvenient edge cases. Industrial biology imposes the same distinction with less forgiving materials. A strain that produces once is a demonstration. A process that produces to specification, batch after batch, begins to look like a product. The distance between those states is where process knowledge accumulates and where a patent portfolio meets operating craft.
It also changes how one should read BIOMEDICAN’s compact team. A small biotechnology company cannot brute-force every question. It must decide which biological variables deserve attention, which measurements will shorten the next experiment, and which steps belong with outside manufacturing partners. The founder’s job is partly scientific judgment and partly the design of an organization that learns faster than its cash disappears. In that sense, Mikheev is still building an agent-based model. The agents now wear lab coats, negotiate contracts and occasionally clog a filter.
The organism is not just a vessel. Choose it well, and its natural habits become part of the business model.
A founder fluent in three languages
Mikheev lists Russian and English on his public profile, but his professional fluency is more usefully counted in three other languages: laboratory biology, computational abstraction and startup economics. Each community has a different standard of proof. A scientist asks whether the experiment survives controls. An engineer asks whether the process repeats. An investor asks whether it scales into a market. A biotechnology founder must answer all three without confusing them.
His public manner leans toward connection. Mikheev describes himself on LinkedIn as an open networker, and his résumé is densely collaborative: long author lists, joint patents, cross-institutional projects, co-founded companies. The solitary-genius portrait never fitted modern biology anyway. A usable strain is a stack of contributions, and a manufacturing process is a relay race disguised as a flowchart.
What remains striking is the continuity. Early in his career, Mikheev worked on tools to identify patterns in viral genomes. In Pittsburgh, he helped model biological behavior as interacting agents. At BioDatomics, he turned genomic analysis into software. At BIOMEDICAN, he is working in the opposite direction: turning information into matter. The sequence travels from molecule to data and back to molecule, with a company wrapped around the return journey.
The elegant version of BIOMEDICAN’s future fits in one sentence: put sugar and an engineered microbe into a vessel, take a valuable compound out. The honest version contains years of strain development, filings, experiments, scale-up and commercial negotiation. Mikheev appears comfortable with both. He knows the sentence is what people remember. He has built a career in the machinery underneath it.