Finsense

AI-powered platform for real-time market tracking, sentiment analysis, fraud detection, and portfolio management — combining FastAPI, WebSockets, and Gemini 2.5 pro for instant financial insights.

Comment

Inspiration

Financial markets operate at lightning speed, yet most AI systems lag behind, relying on outdated or static data. Our team was inspired by this gap — the need for a platform where AI reacts instantly to new information. We envisioned FinSense AI, an adaptive system that continuously learns from live market data, ensuring no critical signal is ever missed.

What it does

FinSense AI is a real-time financial intelligence platform that integrates live data streams, AI reasoning, and natural language processing. It performs market analysis, portfolio tracking, fraud detection, and sentiment analysis — all in one interface. The system provides traders, analysts, and compliance officers with actionable, up-to-date insights powered by continuous data awareness.

How we built it

We built the backend using FastAPI and Pathway, enabling low-latency data streaming and stateful computation. Trae AI orchestrates event-driven AI pipelines, routing financial signals, market data, and news updates into inference modules in real time. The frontend, built with Next.js 14, TypeScript, and Tailwind CSS, provides dynamic dashboards for visualization. We integrated OpenAI GPT-4 and Google Gemini for reasoning, report generation, and anomaly explanation, supported by custom NLP models for sentiment scoring.

Challenges we ran into

Maintaining consistent latency while streaming data from multiple APIs. Handling high-frequency data ingestion without performance drops. Optimizing Trae AI workflows and Pathway pipelines under heavy load. Designing an intuitive UI that keeps complex analytics readable and responsive.

Accomplishments that we're proud of

Built a fully functional real-time financial intelligence system. Achieved sub-second latency using Pathway’s event streaming. Integrated multiple AI models to enhance financial reasoning and sentiment accuracy. Designed a clean, scalable frontend that adapts seamlessly to dynamic data.

What we learned

How to build streaming AI architectures with Pathway and Trae AI. Effective prompt engineering for financial domain reasoning. Techniques for asynchronous data handling in Python. The importance of data visualization for complex financial information.

What's next for FinSense AI – Real-Time Financial Intelligence

Next, we plan to expand FinSense AI with reinforcement learning for adaptive trading strategies and integrate real-time voice assistants for instant market briefings. We also aim to deploy the platform on cloud-native microservices, scale multi-user access, and explore compliance-grade AI auditing to make FinSense AI enterprise-ready.

Built With

Share this project:

Updates