I am an inexperienced idiot. Here's some embarrassing info about me

Parth

22M, CS @IIIT'25. Solo-levelling. Open for collaboration

— Currently participating in AIMO3 (currently ranked #1)
— Obsessed with Bach’s cantatas
— Favourite reads: Gödel, Escher, Bach; A Man’s Search for Meaning; Understanding
— Cinephile: Utopia (uk), Westworld (S1), Burning, Perfect Days, more on my Letterboxd

Projects

Logorrhea: Multi-Agent AI Debate System
Manya: Multi-Agent AI Manim Generation System
DiffusionLM: Diffusion Language Model from scratch
GANDHI: Generative Adversarial Network for Decoding High-level Images from EEG signals
DaVinki: Photomosaic Generator using CIFAR-10 Dataset
BabyDriver: Real Time Vehicle Detection & Speed Estimation

Publications

Multi-Modal Heterogeneous Graph Attention Networks with Dynamic Edge Learning and Cross-Attention Fusion for Fake News Detection

International Conference on Pattern Recognition and Machine Intelligence (Main Track)

Achievements

Kaggle Competitions Expert - Highest Ranked Top 300 in global Kaggle Competitions
Also Notebook Master (Highest Rank: 30) and Dataset Expert
AI Maths Olympiad 3 - Currently Ranked #1
Longest Leader Prize Winner uses my open source notebook
Kaggle Playground Series Competition – Season 13 Episode 18 - Winner (1^st place)
Implemented an ensemble pipeline using feature engineering, GMM clustering, and gradient boosting models
Google – GenAI Exchange Hackathon for Sarvam AI Track - Finalist (Top 5)
Created a simple course creation co-pilot with Indic language translation support
Konwinski Prize - Silver Medalist (Ranked 37^th/ 617)
Solved SWE-bench based GitHub issues benchmark using DeepSeek (got lucky on the leaderboard)
ARC Prize Competition 2025 - Silver Medalist (Ranked 62^nd/ 1455)
Combined domain-specific language approaches with Qwen LLM for solving abstract reasoning tasks
Stanford RNA 3D Folding - Silver Medalist (Ranked 65^th/ 1516)
Utilized DRfold, an RNA 3D structure prediction method to predict RNA backbone coordinates, along with energy-based scoring for structure selection
NeurIPS 2024 Competition: Predict New Medicines with BELKA - Silver Medalist (Ranked 73^rd/ 1946)
Trained a 1D CNN model for molecular data using SMILES representations on TPU
Child Mind Institute Competition: Problematic Internet Use Competition - Silver Medalist (Ranked 74^th/ 3628)
Applied a simple LightGBM model with stratified cv for behavioral prediction
ARC Prize Competition 2024 - Bronze Medalist (Ranked 111^th/ 1427)
Combined domain-specific language approaches with Qwen LLM for solving abstract reasoning tasks
Maastricht University Competition: Game-Playing Strength of MCTS Variants Competition - Bronze Medalist (Ranked 134^th/ 1608)
Ensembled public solutions with LightGBM and CatBoost using preprocessing optimization (didn't do much)
CMI Competition - Detect Behavior with Sensor Data - Bronze Medalist (Ranked 256^th/ 2657)
Implemented a custom multimodal model splitting & fusing IMU, thermal, ToF sensor data with LSTM/GRU
Amazon – ML Challenge - Secured 263^rd place out of 25000 participants (Top 2%)
Extracted product information from images using TrOCR and PaddleOCR models
Rohlik Orders Forecasting Challenge - Ranked 42^nd out of 1017 participants (Top 4%)
Forecasted customer orders using an ensemble of eight traditional regression models
BrisT1D Blood Glucose Prediction Competition - Top 100
Developed blood glucose prediction models for diabetes management research (had a solution which could have place me in top 20 , but did not select)
Optiver – Trading at the Close - Top 10%
Built a trading algorithm booster model with feature engineering for stock price movements prediction
Subway Surfers - Champion
Ranked 1 Champion in Week 47