Leobit
Chemistry Experiment Management Platform
End-to-end MVP development from discovery to deployment that enables researchers to document, analyze, and collaborate on experiments in a structured environment
ABOUT the project
- Client:
- Chemical supplier and integrated drug discovery organization
- Location:
-
USA
|
UA
- Company Size:
- 1,001+ Employees
- Industry:
-
Farmaceutical/Chemical Manufacturing
- Solution:
-
Web and Mobile Development
Technologies:
Leobit helped a global integrated drug discovery organization to build a digital laboratory notebook available as a web and mobile application. The platform is designed for researchers who need a structured way to create, manage, and share experimental data. The first release introduced the core workflows researchers need every day. Users could onboard, create experiments from scratch, or generate them by scanning documents using OCR and NLP. They could edit and organize their work, share experiments, and collaborate with others. After the MVP, the project moved into a Post-MVP phase. The focus shifted to more advanced features, deeper domain logic, and adding automated calculations to support scientific workflows.
Leobit has great technical expertise. They are very proactive and often suggest ways to make the product better. The team managed a complex project and delivered high-quality development. The platform now combines OCR, NLP, and collaboration features with a consistent experience on both web and mobile. They communicated clearly and proved to be a highly reliable partner throughout the project.
Customer
Our customer is a global leader in chemical building blocks, screening compounds, compound libraries, and drug‑discovery services, founded in 1991 and operating globally with chemistry divisions in Ukraine, Germany, and Latvia, and a biology/HTS branch in Poland.
Business Challenge
The client needed a modern digital platform to replace traditional paper-based lab notebooks. Their teams were working across different locations, which made it difficult to maintain consistent documentation and share results efficiently. They wanted a centralized system that allowed researchers to create, store, and collaborate on experiments in a structured format. At the same time, the platform needed to support scientific accuracy, reproducibility, and advanced chemistry-specific features. The challenge was to deliver all of this in a way that was scalable, secure, and cost-efficient from the start.
Why Leobit
The client chose Leobit for its ability to cover the full product development cycle and move quickly from idea to working solution. From the early discovery phase through design, development, and deployment, Leobit provided a single, coordinated team that could handle both product and technical decisions.
Project
in detail
Leobit designed and built a cloud-based platform on Microsoft Azure, structured around modular services and clear separation of responsibilities. The team focused on delivering a working MVP quickly while laying the groundwork for scalability, security, and future AI-driven features.
Leobit developed the core application layer that allows researchers to create, edit, and manage experiments in a structured digital format. The team built a .NET 9 back-end API to handle business logic and data operations, paired with a React-based front end for a responsive user experience. Users can create experiments from scratch, organize their work, and collaborate with team members in real time.
Leobit built AI-driven microservices to automate the creation of experiments from unstructured inputs. Using Python and Flask, the team implemented services that process scanned documents through OCR and NLP pipelines. The system extracts text, key-value pairs, and structured data from lab documents, which significantly reduces manual input. Integration with Azure AI services enables reliable document parsing and supports future expansion into more advanced AI features.
Leobit ensured that sensitive data is never exposed in source code. The team configured all secrets, including database credentials and API keys, to be injected securely at runtime through environment variables in Azure App Service. This approach reduces risk and simplifies secret rotation without requiring code changes.
We set up automated pipelines to streamline development and deployment. Using Azure DevOps and GitHub Actions, the team enabled continuous integration and delivery across all services. Container images are stored in Azure Container Registry and GitHub Container Registry, ensuring consistent builds and reliable deployments across environments.
Leobit implemented centralized monitoring to maintain system visibility and performance. The team configured Azure Log Analytics to collect infrastructure and container logs, while Application Insights tracks application performance. We also introduced cost management controls, including budget alerts and log ingestion caps, to prevent unexpected spending and keep the platform within predictable limits.
We designed a dual-database approach to balance reliability and flexibility. Our team used Azure-managed PostgreSQL for core application data, ensuring automated backups and minimal maintenance. For the RDKit environment, we implemented scheduled backups using cron jobs that generate database dumps and store them in Azure Blob Storage. This ensures data safety even in a self-managed setup.
Leobit built the platform with resilience in mind. By running stateless Docker containers, the team ensured that services can be quickly redeployed in case of failure. This design minimizes downtime and allows the system to recover quickly without complex recovery procedures.
Product discovery phase
Leobit led a structured product discovery phase through a series of focused workshops that helped define the MVP scope. The team facilitated sessions covering scope definition, feature decomposition, and deep dives into template structures to ensure the product matched real research workflows.
Leobit also guided discussions around admin capabilities, collaboration features, and key user journeys. As the workshops progressed, the team refined experiment flows in detail and identified how AI features would integrate into the platform.
Given the need to move efficiently and keep the MVP lean, Leobit’s tech lead also recommended using Stream Chat .NET for real-time messaging instead of building this functionality from scratch. The library provides ready-to-use messaging infrastructure, including real-time communication, message history, scalability, and built-in security features, allowing the team to focus on core product functionality.
The outcomes of the product discovery phase were wireframes that visualize the experience, work breakdown structure document and described in detail calculations for the experiment page.
UI/UX Design
Leobit adapted the design process to fit a fast-moving MVP timeline while still delivering a usable and scalable interface. Since the overall budget was focused on delivering an MVP efficiently, the team prioritized which functionality to include in each release. This planning was led by the Business Analyst and ensured that design efforts stayed aligned with the most critical features.
The client came well-prepared with a defined brand book, user personas, and competitor analysis. As an active end user of the product, the client had a clear understanding of real workflows and constraints. This allowed Leobit to move directly into structuring key user journeys and functionality. Leobit involved a designer early during the discovery phase, working closely with the Business Analyst on user story mapping and feature prioritization. Once the MVP scope was finalized, the designer focused first on the most complex flows with dense data and interactions. This ensured that the hardest UX problems were solved upfront.
Leobit also developed a lightweight design system alongside UI implementation. This created consistency across the interface and made it easier to scale the product in future releases without reworking the design from scratch.
Chemical structure search & RDKit integration
Leobit implemented chemistry-specific capabilities by integrating RDKit into a dedicated PostgreSQL environment. The team deployed a custom virtual machine running PostgreSQL with RDKit extensions to support advanced chemical queries. This setup enables exact-match, substructure, and similarity searches, enabling researchers to work with chemical data in a meaningful way. The architecture separates this workload from the main database to maintain performance and flexibility.
Scalable microservices and containerized infrastructure
Leobit built the platform using Docker containers to ensure consistency across environments and simplify deployment. All services run on a shared Linux App Service Plan, balancing performance with cost efficiency. The team adopted a microservices approach to isolate core application logic from AI services. This makes it easier to scale components independently and introduce new features without disrupting the system.
Leobit designed the infrastructure with security as a priority. The team implemented a Virtual Network that routes all internal communication through private channels, minimizing exposure to the public internet. We configured access restrictions to ensure that only authorized services and users can reach critical components. The chatbot and ML services operate behind controlled endpoints, and database access is limited to private IP ranges.
UI system development
Leobit built a scalable and consistent UI system to support fast development and a clean user experience across the platform. The team used Radix UI primitives as the foundation for accessible, unstyled components such as dialogs, tabs, selects, and tooltips. This allowed full control over design while maintaining accessibility standards.
To implement styling, Leobit adopted Tailwind CSS, enabling rapid UI development with a consistent design language. This approach made it easy to maintain visual coherence across screens and iterate quickly during the MVP phase.
To capture user feedback and interactions, the team integrated Sonner to display toast notifications, ensuring users receive clear, real-time system responses. Data-heavy views were powered by TanStack Table, which provides built-in support for sorting, pagination, and efficient data handling. Together, these tools allowed Leobit to deliver a responsive, user-friendly interface that scales with the product and supports complex research workflows.
Technology Solutions
- Used Azure AI Services (Document Intelligence, OpenAI, Speech-to-Text) to power document parsing, text extraction, and AI-driven features.
- Integrated PostgreSQL + RDKit (VM-based) for chemistry-specific processing, since RDKit enables advanced chemical structure queries that standard databases cannot handle.
- Integrated Azure Blob Storage for storing files such as document uploads and database backups.
- Applied TanStack Table for handling data tables and supporting sorting, pagination, and efficient rendering of large datasets.
- Used Ketcher as a chemical structure editor to allow users to draw and interact with molecular structures directly in the UI.
Value Delivered
- Delivered a fully functional MVP in under 3 months, enabling the client to validate the product quickly with real users.
- Replaced manual, paper-based lab workflows with a structured digital system, improving data consistency and traceability.
- Reduced experiment documentation time through AI-powered OCR and NLP automation.
- Enabled advanced chemical structure search (exact, substructure, similarity) using RDKit integration.
