Final-year engineering student who builds things at the intersection of machine learning, full-stack web, and mobile development. This is where I put the stuff I make.
A Flutter app that aggregates news from reputed sources and runs each article through a TensorFlow ML model to detect fake news. Keeps feeds fresh via automated GitHub Actions, and also surfaces daily weather and live currency exchange rates alongside the headlines.
A dual-layer AI architecture that simulates emotion by separating cognition into a fast "unconscious" system and a slower, reflective "conscious" layer. The unconscious layer generates instinctive responses; the conscious layer evaluates, suppresses, or amplifies them - mimicking how emotion and reason interact in human cognition.
A benchmark that evaluates how well LLMs handle real-world OCR output - corrupted text, missing elements, garbled structure and all. Models are scored on their ability to parse noisy OCR input and produce clean, corrected JSON, testing a capability that matters in production but is absent from standard evals.
A two-player online game based on the board game played by Columbina in Genshin Impact - a special variant of tic-tac-toe with its own ruleset. Players create a room, share a code, and play in real-time; Firebase handles matchmaking, game state sync, and score tracking across sessions.
Point your camera at any document and Docsum extracts the text via OCR and returns a clean summary. Built in Flutter with a Python summarisation API on the backend - useful for quickly digesting printed reports, receipts, or handwritten notes without typing a word.
I'm Ashwin Shetty, a final-year engineering student graduating in 2026. I build things across the full stack - mobile apps, web platforms, machine learning systems, and the occasional AI research tool.
My work ranges from practical (a camera-to-summary document tool, a fake news detector) to exploratory - like a dual-layer cognitive architecture that models how emotion and reason interact in AI systems, or a benchmark for how well LLMs handle messy real-world OCR input.
Outside of code, I enjoy reading stories - especially webnovels.
Built a modular cloud dashboard for managing security camera feeds and analytics. Integrated AWS Amplify and S3 for authentication and secure storage. Designed camera grid views and analytics features using a plugin-based architecture.
Worked on data analysis and visualisation for machine learning models. Built a social media sentiment analysis system using a bidirectional LSTM RNN.
Relevant coursework: Machine Learning, Data Structures & Algorithms, Database Systems, Communication Networks, Information Security, Operating Systems, Cloud Computing.
Awarded a ₹1,00,000 scholarship for academic excellence.
If something here caught your eye or you just want to say hi - feel free to reach out. Always happy to talk about interesting ideas, projects, or anything I've built.