what's cooking in my mind
- developing transformer architectures for time series forecasting
- exploring few-shot learning techniques
- building multimodal vision-language models
- researching efficient model compression methods
some fun projects i am working on:
prompt banana: tool to help with prompt engineering for neo banana and veo3 models.
guess the subreddit: classification model for guessing the subreddit based on the title of a post.
i'm hoping to spend summer working alongside defi research teams to deepen my understanding of onchain markets.
stuff i like to build
neural style transfer
artistic image transformation
Built a deep learning model that transforms photos into artistic masterpieces using neural style transfer. Combines content and style images to create unique visual compositions with customizable artistic effects.
real-time object detection
computer vision pipeline
Built a real-time object detection system using YOLOv8 that processes video streams with 30+ FPS, featuring custom model training and multi-class detection capabilities.
minigpt ai
open source llm assistant
Lightweight AI assistant built on open source models, featuring natural conversations, code generation, and document analysis capabilities.
Solana Screener
browser extension for defi traders
Chrome extension that instantly converts Solana token addresses to Dexscreener links for quick analysis. Built for traders who need fast access to token data.
Text Behind Pics
dynamic content positioning system
Advanced webapp for intelligently positioning text and image overlays behind images. Uses background masking models to detect optimal placement zones and ensures readability while maintaining visual hierarchy.
Poly Watcher
prediction market analysis platform
Multiagent AI system that provides deep analysis reports and probabilistic assessments for prediction markets. Combines multiple specialized agents for comprehensive market intelligence and forecasting.
contact
slide into my dms on x/twitter if you're cooking up something spicy 🫡