Accessing Clean Financial Data with Upstox's Company Fundamentals API

🚀 Excited to share my latest project: InvestSense AI I built a full-stack fintech portfolio management platform that combines AI agents with real-time market data — here's what's under the hood: 🔧 Tech Stack: Backend: FastAPI + MongoDB, deployed on Render Frontend: React/Vite, deployed on Netlify AI Agent: LangGraph orchestration powered by Groq's Llama 3.3 70B Auth: JWT-based secure authentication Market Data: Live NSE (Indian stock market) data via direct Yahoo Finance chart API integration 💡 The Challenge: Getting reliable, real-time Indian stock data turned out to be trickier than expected — both yfinance and Twelve Data failed to deliver consistent results for NSE stocks. I ended up building a direct integration with Yahoo Finance's chart API to solve this, which taught me a lot about working around third-party API limitations in production. 📊 What it does: InvestSense AI helps users manage their stock portfolios with an AI agent that can reason over live market data, answer portfolio-related queries, and assist with investment decisions — all wrapped in a clean, responsive interface. This project pushed me deeper into agentic AI systems (LangGraph), production-grade API design, and the real-world messiness of integrating financial data sources. Always learning, always building. 🔨 🔗 Check it out: investsensepro.netlify.app #AI #FinTech #LangGraph #FastAPI #GenAI #Groq #BuildInPublic #AIEngineering

Just launched: Algo Pilot Basic @ $19/month A lower-cost way to build and run automated trading strategies without coding. Basic includes access to the AI builder tool, hourly & daily trading, 250 live trades per month, and a 2-year backtest window. We’ve also upgraded existing plans with zero price increase: • Standard plan gets up to 500 live trades/month (double the previous) • Ultimate plan now has Unlimited live trades And our new AI Builder Tool is now in Beta for all users. Describe your idea in plain English and get a customizable algorithm draft you can backtest immediately. AI Builder post: https://lnkd.in/eDSbBMAZ Making algorithmic trading more accessible one update at a time. #AlgoTrading #FinTech #AlgorithmicTrading

Today marks an important milestone for me. Over the past few months, I've been building Aureon AI Trading Coach, the first product within the Aureon Capital AI ecosystem. What started as an idea for a better trading journal evolved into something much bigger. While building it, I kept asking one question: Why do most trading tools tell us what happened, but not why it happened? That question shaped the architecture of Aureon. Instead of only analyzing broker data, it captures the trader's reasoning through structured trade reviews, builds Trader Memory, identifies recurring patterns, and only declares a trading edge when there's enough evidence to support it. Today, I'm proud to release Aureon AI Trading Coach v1.0.0 as an open-source project. This is only the first step. The long-term vision is Aureon Capital AI—an ecosystem of AI-powered products built to help people make better decisions in financial markets. I'm grateful for everything I've learned while building this first version, and I'm looking forward to the feedback that will shape what's next. Turning Decisions Into Intelligence. 👇 You can read the full announcement below.

Introducing Aureon AI Trading Coach v1.0.0 Today marks an important milestone for Aureon Capital AI with the first stable release of Aureon AI Trading Coach. Aureon AI Trading Coach is an open-source trading intelligence system designed to help traders move beyond recording trades and begin understanding the decisions behind them. Unlike traditional trading journals that focus primarily on historical statistics, Aureon combines structured trade reviews with performance analytics, Trader Memory, Pattern Discovery, Edge Discovery, and AI-assisted coaching to build a richer understanding of trading behaviour over time. Version 1.0.0 includes: ✅ Trade import from MetaTrader 5 and Exness ✅ Performance analytics and risk metrics ✅ Structured post-trade reviews ✅ Decision Records ✅ Trader Memory ✅ Pattern Discovery ✅ Edge Discovery ✅ AI-powered trading coach ✅ FastAPI REST API ✅ Model Context Protocol (MCP) integration The release establishes the foundation for the broader Aureon Capital AI ecosystem, where intelligent software is designed to help traders learn from their decisions through evidence-based analysis rather than assumptions. This is only the beginning. Future releases will continue expanding Aureon's capabilities with deeper intelligence, broader integrations, richer analytics, and enhanced AI coaching. Turning Decisions Into Intelligence. Explore the project on GitHub: 🔗 https://lnkd.in/dtg-hutV #AureonCapitalAI #ArtificialIntelligence #Trading #OpenSource #FinTech #Python #FastAPI #SoftwareEngineering

I've been building StockSense AI — an investment research tool for NSE and major US stocks that tries to do something most stock apps don't: actually explain why a stock scores the way it does, in plain English. It scores every stock across six independent angles — fundamentals, technicals, news sentiment, sector strength, macro conditions, and geopolitical risk — then synthesizes all of that into a real research narrative, not just a number. You get an Opportunity Score, a Confidence Score (how much to trust it), and a Risk Score, each grounded in the actual underlying data, not a black box. Also has portfolio tracking, a watchlist, and a screener — free to browse and search, no account needed for the basics. Still very much a work in progress (solo project, built end to end — backend, frontend, infra, deployment, all of it), but it's live and I'd genuinely love feedback from anyone who tracks stocks or is curious about the space. 🔗 https://lnkd.in/dCsUQWeG #buildinpublic #fintech #investing #stockmarket

AI test generation looks free: one toggle, and coverage jumps forty points overnight. The bill arrives later, and it isn't denominated in money. The hidden cost is trust debt. Generated tests that assert nothing meaningful inflate the coverage number while verifying little. Tests generated from buggy code lock the bugs in place, so the eventual fix reads as a regression. Flaky tests pile up until the team learns to ignore red, and a twenty-minute CI run teaches everyone to skip the suite exactly when deadlines bite. Six months in, you own a very green dashboard that nobody on the team actually believes. The cost-control is human and cheap relative to the bill: generated assertions get reviewed against the spec, not the code, regression suites go first because there current behavior genuinely is the spec, coverage gates in CI keep the bar honest, and flaky tests get deleted the week they flake instead of muted and forgotten. Paid up front, the review cost is hours. Paid later, the trust debt is incidents, rework, and a suite the team quietly routes around. How does your team handle this? Book a Free Call → devxhub.com #AIDevelopment #AIEngineering #SoftwareDevelopment #AICoding #TechTrends2026 #Devxhub

I recently submitted FundersAI to OpenAI Build Week 🚀 FundersAI is an AI-powered research platform designed to reduce the friction involved in researching mutual funds and financial markets. The platform brings together: • LLM-powered research workflows • Retrieval-Augmented Generation (RAG) • Structured financial data processing • Document and portfolio analysis • FastAPI backend services • Next.js and Supabase • Automated data ingestion pipelines Building FundersAI has helped me work through real production challenges such as data reliability, retrieval quality, model cost, latency, observability, and integrating AI features into a usable product. The submission gave me an opportunity to refine the product, improve its AI workflows, and think more deeply about how GenAI can make financial research more accessible without presenting itself as investment advice. I am still actively developing FundersAI and would appreciate feedback from people working in AI, fintech, and software engineering. 🔗 Explore FundersAI: https://fundersai.co.in 💻 GitHub repository: https://lnkd.in/dK8VAuqA #OpenAIBuildWeek #GenerativeAI #MachineLearning #RAG #LLM #AgenticAI #FinTech #FastAPI #NextJS #BuildInPublic #OpenAIBuildWeek #GenerativeAI #MachineLearning #RAG #LLM #AgenticAI #FinTech #FastAPI #NextJS #BuildInPublic

🚀 Excited to share FinTrack AI, an AI-powered financial research and trading journal platform. This project combines live market data with AI-generated insights to help users research financial assets, maintain a trading journal, and analyze trading performance. Through this project, I gained hands-on experience with AI integration, REST APIs, authentication, and building a complete application from idea to deployment. 🔗 GitHub: https://lnkd.in/gQSgkqJ7 🎥 Demo: https://lnkd.in/gz-iiVfp I'm always open to feedback and suggestions! #AI #MachineLearning #ReactJS #NodeJS #FinTech

I just shipped something I'm genuinely proud of. 🚀 Just wrapped up the Be10X AI Generalist Hackathon, and I built WealthPulse AI to solve the pain of manual financial tracking! 🌐 Try the Live App: https://lnkd.in/ePWtp3iA The Pain: Parsing messy bank and credit card PDF statements across multiple accounts requires 3 to 4 hours of tedious manual spreadsheet data entry every month. The Solution: WealthPulse AI automates this completely, turning hours of manual work into less than 10 seconds of automated financial intelligence. Under the Hood: I built the platform using React, Node.js, Vercel, Render, and Google's Gemini 3.1 Flash Lite model. The AI ingests unstructured PDF bank statements, categorizes transactions into groups like EMI, Medical, and Groceries, flags spending anomalies, and populates a clean, interactive dashboard. Biggest Challenge? Handling dense, multi-page PDF statements simultaneously without hitting AI rate limits. Strategically refactoring the backend logic to integrate Gemini 3.1 Flash Lite provided the high throughput and low latency needed for instant document extraction. Biggest Takeaway? How easily modern AI can convert raw, unstructured document text into structured, actionable insights that save hours of human effort. Huge thanks to the team at Be10x for organising a hackathon that pushes developers to ship practical, real-world solutions! #AI #Hackathon #Be10X #BuildInPublic #WealthPulseAI #FinTech #FullStack #ReactJS #NodeJS #NoCode

A week ago, we launched VENTY AI ( venty.72street.ai ), our AI powered stock market researcher. As the CTO & Co-founder of 72 Street, I wanted to share the thinking behind the name because it captures something fundamental about what we are building. Our brand is 72 Street. 72 = Say VENTY to. Say VENTY to analyse Reliance. Say VENTY to check the PE ratio of TCS. Say VENTY to understand what is happening with your stocks. Every time someone says VENTY, they are unknowingly saying 72 Street. But for me, the more interesting part is what sits behind those words. Building VENTY was not about putting a chatbot on top of market data. The real challenge was making AI work with large amounts of financial information and turn it into useful, contextual research. VENTY scans 3,800+ NSE stocks, brings together fundamental, technical and sentiment analysis, and helps turn complex market information into answers that are easier to understand. The goal was simple. Make the speed of AI work alongside the depth of serious research. Getting that balance right took a lot of iteration, testing and engineering. We are still improving it every day. A week in, seeing people actually say “VENTY” while researching the market has been incredibly rewarding. This is just the beginning. Say VENTY to the market. #VentyAI #72Street #ArtificialIntelligence #StockMarket #Fintech