MarketMurmur
MarketMurmur is a market intelligence platform where you can search any public company and instantly get a clear, all in one view of what's going on. It brings together stock price trends, key financial metrics, and recent news sentiment into a single dashboard.
Under the hood, it runs automated data pipelines using n8n to handle continuous data ingestion, transformation, and storage of stock data, financials, and news into a centralized PostgreSQL database. Each news article is processed with setiment analysis, and all signals are normalized and combined into a Hype vs Fundamentals score that flags companies as overhyped, undervalued, or aligned.
F1 Race Intelligence
A race analytics platform to explore Formula 1 races, from lap times and tyre strategy to pit stops and results, all in one place. It also generates AI driven strategy recommendations and race commentary based on real race scenarios
Built on a data pipeline powered by Kafka and Spark, the system processes race data into a feature rich dataset that feeds a RAG pipeline using Pinecone and CrewAI. This allows the app to retrieve similar historical situations and generate context aware decisions, giving users a smarter way to understand race strategy and performance.
Pennywise
An expense tracker built with React Native that lets users log spending naturally, just by typing it like a text message.
Behind the scenes, it runs a text-based expense ingestion pipeline using NLP techniques, including Named Entity Recognition (NER), to extract key details like amount and item and convert them into structured transaction records for automated categorization and spending analysis.
The app then compiles everything into clear summaries and insights, helping users understand patterns in their spending and make better budgeting decisions without manual effort.
SubTrack
This app makes it easier to keep track of recurring subscriptions. Users can choose from common services or add their own, set reminders, and fully customize billing cycles down to any duration they want.
The focus was on keeping the interface clean and simple, so everything from viewing payment schedules to managing multiple subscriptions feels straightforward and easy to use.
PWagon
A deep learning system for real time license plate detection that lets you post a "wanted vehicle" and instantly start tracking it across a network of cameras. Raspberry Pi devices scan live feeds, detect plates using a convolutional neural network trained on a custom image dataset, and use Google OCR to extract text and log sightings with time and location.
Deployed for Amber Alert style workflows, it turns simple cameras into an automated monitoring system where alerts trigger instantly and matches are detected in real time, removing the need for manual footage review and enabling fast response when it matters most.
Jasper.AI
Jasper.AI is an interview prep platform that builds a fully personalized roadmap based on your target company, role, timeline, and daily availability so no two candidates get the same plan.
It uses LLM pipelines to research real interview questions from Reddit, Glassdoor, and LeetCode, prioritizes high impact weak areas using your confidence scores, and guides you day by day. You can practice through live voice-based mock interviews, where a feedback loop evaluates your responses and iteratively refines prompts based on adaptive assessment, providing structured, scored feedback.
THE INVISIBLE PATIENT
An AI powered mental health companion for caregivers that offers voice or text check-ins. It runs a ZBI based burnout assessment naturally through conversation and tracks emotional patterns using a proprietary Resonance Score.
It detects crisis signals in real time and includes an anonymous peer support forum. All data is handled in a privacy first, zero friction experience that helps catch caregiver burnout early.
