Building intelligent analysis systems for one of Sweden's leading bioinformatics companies — and the global marketplace that brings their discoveries to market.
Cancer research is, at its core, a data problem. Clinical trial results, patient case narratives, genomic datasets, and multi-source biological records each carry meaningful signal — but the volume and heterogeneity of that data make consistent, thorough analysis genuinely difficult. Under the pressures of active research, subtle patterns get missed. Insights that should surface in weeks take months. Connections across datasets that could change a hypothesis go unnoticed.
Bioseeker's mission is to turn this complexity into usable intelligence — the kind that medical professionals, pharmaceutical researchers, and clinicians can act on. To do that at scale, they needed more than better workflows. They needed systems that could think alongside their analysts: ingesting diverse inputs, identifying non-obvious relationships, and producing insights with the reliability and depth that oncology research demands.
Two parallel challenges required simultaneous attention: the intelligence layer — ML agents capable of processing vast, multi-modal research datasets — and the distribution layer, a production-grade global marketplace to bring Bioseeker's bioinformatics reports and business intelligence to the researchers and institutions that need them.
Bioinformatics at this level means working with data that is simultaneously enormous in volume, heterogeneous in format, and deeply consequential in interpretation. Clinical trial results and patient narratives don't share a schema. The patterns that matter most are often the ones conventional analysis misses under high workload. Building systems that handle this reliably requires both ML expertise and a genuine understanding of the research context.
We operated as both System Developer and Project Lead across the engagement — holding responsibility for technical architecture and delivery while keeping the broader team aligned and the project moving.
The core of the work was designing and building ML agents trained to process diverse cancer research inputs. These systems handled clinical trial results, patient narrative data, and multi-dimensional biological datasets — not as isolated pipelines, but as integrated feeds informing a shared analytical model. The agents were built to surface subtle patterns that conventional review processes miss under high workloads: correlations across patient cohorts, anomalies in trial data, signals in unstructured narratives that point toward mechanisms or risk factors worth investigating.
The design emphasis was on breadth of input and reliability of output. Researchers needed to trust what the system surfaced — which meant rigorous validation, clear confidence signalling, and outputs that fit naturally into existing research workflows rather than demanding that researchers adapt to the tool.
In parallel, we led development and ongoing maintenance of the Biomarket Group e-commerce platform — Bioseeker's global distribution channel for business intelligence and bioinformatics reports. This was not a standard web commerce build: the platform managed proprietary scientific content with access controls for institutional buyers, handled licensing across jurisdictions, and needed to operate reliably for a global market of research institutions, pharmaceutical companies, and clinical organisations.
The ML agents deployed into Bioseeker's research workflows materially changed the pace and depth of analysis — processing inputs at a scale and consistency that human review under high workload cannot match.
Tobias Salem joined our team at a strategically important time and significantly improved the technological platform of the company — not only in practice but also in vision for future development. As technical director, he oversaw operations from BioSeeker's business modelling of the pharmaceutical industry to our e-commerce solutions. Always appreciated among his co-workers, always constructive in any discussion, and truly instrumental to my present understanding of what information technology can achieve.
I had the privilege to work with Tobias as a developer in BioSeeker Group. Tobias is a very skilled and focused developer and systems designer, who always has overview over the objectives and overall software design and gets the job done in time. I learned a lot about software design from Tobias and I will always remember what a good and cheerful company he was in the office.
The stakes in oncology research are unlike those in almost any other domain. The data being analysed — clinical trials, patient outcomes, biological mechanisms — represents real people and real treatment decisions. A pattern missed because an analyst was overwhelmed is not a reporting gap. It is a hypothesis that never formed, a correlation that never informed a trial, a signal that never reached the clinician who needed it.
Building ML systems for this context demands more than technical capability. It requires a genuine understanding of what researchers need from their tools — reliability, interpretability, and outputs that integrate naturally into the science rather than sitting alongside it. The systems we built for Bioseeker were designed with that understanding at the centre: not just to process data faster, but to help the people doing the research do better science.
This is the approach we bring to every engagement where the output matters beyond the screen: engineering precision in service of a mission that is worth getting right.
If you're working with complex data in life sciences, healthcare, or research — we want to understand the problem.
Start a conversation