Software has transformed how we communicate, collaborate, and run businesses. Increasingly, software engineers are turning their attention to the physical world, asking: what are the equivalent problems in science, energy, or manufacturing where my skills could have a massive impact? 

Developing new medicines is the most effective way to improve human health and reduce medical costs. But unlike software, there isn’t a rich stack of tools to manage the development process. Software engineering is agile, sprint driven, and iterative. You break work into tickets, ship code, test with users, and refactor when something breaks. The loops are short, the costs of being wrong are low, and if a release goes sideways, you can usually roll back and try again.

Biotech R&D is nothing like that. Developing a new medicine is closer to launching a rocket into space: path dependent, capital intensive, and defined by critical milestones that determine success or failure. Even in the best case, moving a discovery into the clinic takes 5-10 years and hundreds of millions of dollars. Add in all the failures along the way, and the costs run into the billions.

The journey from discovery to medicine 

Navigating this complexity is why brilliant science needs to be paired with effective management to develop breakthrough medicines. That starts with a plan.

Source: Clinical and Translational Science

Leaders need to work forward from their technology (what’s possible in the lab) and backward from the patient (what evidence regulators and clinicians require). After defining your strategy, you need to translate that into a multi-year roadmap of scientific activities: assays, animal studies, manufacturing runs, clinical trials. While sprints are a semi ordered list of tasks, this looks more like a directed graph. 

Source: Nature Reviews Drug Discovery

Then you need to layer on a financial model: personnel, equipment, materials, facilities, external partners, all netting out to how much capital you’ll need and when. Even at the earliest stage of company formation the plan can span several years and add up to tens of millions of dollars. Without it, you can’t raise funding.

Plans are useless, but planning is essential

Biotech is defined both by its complexity and its intrinsic uncertainty. Experiments fail. Competitors emerge working on the same target. The FDA asks for more data. Every delay from a CRO, hiccup in manufacturing, or new piece of market intel cascades through the entire roadmap. That makes it critical to continuously track scientific progress and financial spend against the plan.

When things do go wrong, you can’t just patch and redeploy. You have to re-plan, explore new scenarios, and weigh your options all over again. Biotech is a constant, high stakes cycle of planning, execution, and decision making.

Engineering a system to manage complex R&D 

Managing complex R&D means building a system that can plan like a strategist, track like a program manager, read like a scientist, and communicate like an executive. Unlike agile sprints or commercial manufacturing, R&D systems must navigate fundamental uncertainty about timelines, resources, and whether the science will work at all. This creates three technical challenges:

1. Deriving the right primitives to model science, finance, and operations

   Scientific workstreams are graphs expressing study designs, cross team dependencies, and resources. Programs are metagraphs that orchestrate multiple scientific workstreams toward milestones. Biotech companies burn capital traversing these graphs into “value inflection points,” but the work graph and financial model have never been connected. Scientists need to know what they’re working on next week and investors need to know when the company runs out of money. The primitives must be expressive enough for complex scenario planning yet presented intuitively enough that a first time founder can model their entire company.
   
   

2. Building observability into unstructured systems

   In software, you have logs, alerting, and product metrics. In R&D, critical signals are buried in email threads ("manufacturer mentioned QC issues"), meeting transcripts ("toxicity results look unexpected"), and PDF reports. Completing scientific work does not mean that you’ve achieved the desired scientific outcome. We’re using LLMs to extract PhD-level information to assess progress towards scientific milestones. When done right, this gives teams their cognitive bandwidth back to focus on science and strategy instead of constantly gathering information to make decisions.
   
   

3. Scenario modeling and goal based optimization

   R&D’s intrinsic uncertainty means constant scenario planning. Our scenario system works like Git branches: users create a branch of an entity (program, budget) and explore "what if" permutations, then combine scenarios to evaluate complex trade offs. Today, humans manually explore this decision space. Soon, we'll let users express goals ("reach IND by Q3 2026 with 6 months runway") to solve for. Eventually, we want the system to suggest optimal paths through the scenario space. The challenge is building an engine that can generate plausible scenarios, quantify trade offs across multiple dimensions, and present options that scientific leaders can understand and trust.

Help us build the intelligent management platform for R&D 

Our founding team has spent years moving scientific data and workflows from paper to the cloud, but kept seeing brilliant science getting derailed by fragmented planning, missed signals, and ad hoc decision making. We knew that companies needed the same level of software sophistication for managing their R&D programs as they have for managing their scientific data. 

We’ve built the foundation of our product in partnership with dozens of biotechs, and recently announced $20m in funding from Andreessen Horowitz, 8VC, Insight Partners, Definition, and Haystack. Our customers use Orchestra to manage the development of medicines for some of the world’s most pressing problems, such as neurodegeneration, cancer, and metabolic disease.  

We're looking for engineers who thrive in environments where they can shape both technical architecture and product direction. You’ll own features from idea to production, balancing speed with the craftsmanship required for complex, mission critical software. No biotech experience necessary, only intellectual curiosity about how discoveries are developed. If you’re excited about building software to help scientific organizations operate more effectively, please reach out.

Check out our open positions and watch our CEO's recent interview with BiotechTV.