Arnav Goyal
Entrepreneur and Researcher
Based in Bangalore, India
BIOGRAPHY
BIOGRAPHY
Work for impact and build with purpose,
I’m Arnav, an AI, blockchain, and startup builder turning ideas into reality.
Education & Experience
Bachelor of Engineering
Dayananda Sagar Academy of Technology and Management (DSATM)
2024 - 2028
Bachelor of Engineering
Dayananda Sagar Academy of Technology and Management (DSATM)
2024 - 2028
Bachelor of Engineering
Dayananda Sagar Academy of Technology and Management (DSATM)
2024 - 2028
Founder
Jeevni, India
•
2024 - Present
Co-Founder
Event Union, India
•
Oct 2024 - Apr 2025
Young Member
New York Academy of Sciences, New York, USA
•
Sept 2022 - Sept 2024
Member/Coordinator
Bureau of Indian Standards, India
•
July 2022 - May 2024
Co-Founder
Aerospacizm, India
•
Oct 2020 - June 2024
Founder
Jeevni, India
•
2024 - Present
Co-Founder
Event Union, India
•
Oct 2024 - Apr 2025
Young Member
New York Academy of Sciences, New York, USA
•
Sept 2022 - Sept 2024
Member/Coordinator
Bureau of Indian Standards, India
•
July 2022 - May 2024
Co-Founder
Aerospacizm, India
•
Oct 2020 - June 2024
Founder
Jeevni, India
•
2024 - Present
Co-Founder
Event Union, India
•
Oct 2024 - Apr 2025
Young Member
New York Academy of Sciences, New York, USA
•
Sept 2022 - Sept 2024
Member/Coordinator
Bureau of Indian Standards, India
•
July 2022 - May 2024
Co-Founder
Aerospacizm, India
•
Oct 2020 - June 2024
Selected Projects & Work
Neuro-Symbolic & Compositional Reasoning
Solo benchmark comparing an LSTM baseline with a neuro-symbolic pipeline on a synthetic Boolean-expression compositional task
This project shows a clear compositional generalization gap — neural baseline (IID ≈ 0.8713 → compositional ≈ 0.7167) vs symbolic executor (IID = 1.0, compositional = 1.0)
02
Probabilistic ML & Uncertainty
03
CrowdCast — AI-Powered Crowd Density Predictor
Neuro-Symbolic & Compositional Reasoning
Solo benchmark comparing an LSTM baseline with a neuro-symbolic pipeline on a synthetic Boolean-expression compositional task
This project shows a clear compositional generalization gap — neural baseline (IID ≈ 0.8713 → compositional ≈ 0.7167) vs symbolic executor (IID = 1.0, compositional = 1.0)
Probabilistic ML & Uncertainty
Reproducible extension evaluating Bayesian Linear Regression and Gaussian Process baselines with explicit IID vs OOD splits and calibration metrics
Added deterministic pipeline, NLL/RMSE/95% coverage and calibration diagnostics; reported model table (e.g., BLR IID NLL ≈ -0.48, BLR OOD NLL ≈ 2.05; RBF+White kernel shows strong IID NLL/coverage)
CrowdCast — AI-Powered Crowd Density Predictor
Full-stack predictive platform forecasting crowd density (Bengaluru) using web scraping, feature engineering, XGBoost, and an AI assistant layer
Built live data pipeline (Selenium scrapers → feature extraction), XGBoost crowd scoring, interactive geospatial visualizations and Gemini-backed assistant responses
Neuro-Symbolic & Compositional Reasoning
Solo benchmark comparing an LSTM baseline with a neuro-symbolic pipeline on a synthetic Boolean-expression compositional task
This project shows a clear compositional generalization gap — neural baseline (IID ≈ 0.8713 → compositional ≈ 0.7167) vs symbolic executor (IID = 1.0, compositional = 1.0)
Probabilistic ML & Uncertainty
Reproducible extension evaluating Bayesian Linear Regression and Gaussian Process baselines with explicit IID vs OOD splits and calibration metrics
Added deterministic pipeline, NLL/RMSE/95% coverage and calibration diagnostics; reported model table (e.g., BLR IID NLL ≈ -0.48, BLR OOD NLL ≈ 2.05; RBF+White kernel shows strong IID NLL/coverage)
CrowdCast — AI-Powered Crowd Density Predictor
Full-stack predictive platform forecasting crowd density (Bengaluru) using web scraping, feature engineering, XGBoost, and an AI assistant layer
Built live data pipeline (Selenium scrapers → feature extraction), XGBoost crowd scoring, interactive geospatial visualizations and Gemini-backed assistant responses
CoinPlay (blockchain fantasy football)
Blockchain-based fantasy-football prototype (Bitcoin trading via UniSat wallet) — description taken from your CV
End-to-end prototype demonstrating Web3 wallet integration and basic trading/market flows (as listed in your CV)