Context
The client needed a purpose-built networking platform connecting investors with corporate representatives — not a generic social app, but a tool where both sides can discover relevant contacts, schedule live session discussions, share financial documents and access company intelligence.
The platform had to integrate financial data from an international data provider (income statements, balance sheets, cash flow, holders, stock price performance and earnings transcripts), support real-time audio/video live sessions with recording and transcription, and surface intelligent recommendations based on user profiles and preferences.
Solution
Team and project management
- Managed the entire project lifecycle over 1.5 years, including hiring and coordinating a team of over 40 professionals across back-end, front-end, mobile and design.
- Ran the technical coordination end-to-end: architecture decisions, API design sessions with the client, sprint planning and release management.
Architecture
- Designed a microservices architecture (Node.js, Python) deployed on AWS with purpose-specific data stores:
- Neo4j for the social graph — user connections, company relationships and recommendation traversals.
- MongoDB for meeting transcripts and unstructured content.
- SQLite for local replication of financial datasets via SFTP sync.
- Web clients built with React and Vue.js/Nuxt.js; native mobile apps for Android (Kotlin) and iOS (Swift).
- Real-time messaging powered by PubNub; audio/video live sessions via Agora with cloud recording.
Intelligent matching engine
- Built a multi-factor recommendation algorithm that connects investors and corporates based on their profiles and preferences.
- Corporate users see relevant investors, and vice versa — recommendations are driven by overlapping criteria from both sides' profiles.
- Events and live sessions are surfaced based on matching against user profiles and their contact lists.
Financial data integration
- Integrated data feeds from an international financial data provider for company profiles: income statements, balance sheets, cash flow statements, financial ratios, holders and historical stock price performance.
- Synced earnings call transcripts alongside in-app live session transcripts, providing a unified view of company communications.
Automated transcription pipeline
- Built a dedicated service (Python) that automatically processes live session recordings: extracts audio, generates transcripts with per-speaker identification, and stores structured results linked to the meeting and its participants.
- Orchestrated via a message broker between the backend and transcripts service for asynchronous, resilient processing.
Results
- Delivered a production platform with full communication capabilities: messaging, live sessions with recording, file sharing and live document presentation.
- The matching engine enabled investors and corporate representatives to discover relevant contacts they would not have found through manual search.
- Automated transcription eliminated manual note-taking from live sessions and made discussions searchable.
- The microservices architecture allowed independent scaling — the transcripts pipeline and financial data sync run on separate infrastructure without affecting the core platform.