ARYΛN DUTT

Profile

Systems engineer and LLM infrastructure builder. I ship high-throughput models, optimize edge deployments, and write production-grade quant code. Always low-latency. Always high-impact.

Education

• NTU (B.Comp. AI, Math Minor): Algorithms, Probabilistic Models, Neural Nets
• Y Combinator AI School (Invite-Only): Talks w/ Elon, Satya, Sam, Andrew
• IB Diploma: 4 HLs. Founded Java + Rocket Clubs

Execution

• MIT Julia Lab: MoE SVDs. +3.5x speed under GPU constraint
• NTU CCDS: Jailbreak evaluations of LLM safety (arXiv:2404.02151)
• SC3DP: Optimizing DED powder flow with ML
• SEAQR: Mission-critical dashboard for autonomous ops
• Tagit, BioMetallica, ServBridge: AI/CV for fintech, e-waste, enterprise NLP
• RevisionDojo (YC): Backend infra for scraping & structuring datasets

Projects

• SmartScanGPT: OCR-to-whitepaper agent. Offline, optimized. GitHub
• Portfolio Opt: Stress-tested financial simulator + AI forecasts GitHub
• Causal Router: MoE inference w/ counterfactual logic GitHub
• HyperRoute: IRS/MA routing sim for NYSE-SGX under 70ms GitHub

Stack

Code: Python, C++, Java, Julia, SQL, Bash, JavaScript

ML: PyTorch, JAX, OpenCV, GPT APIs, NumPy, Pandas

Finance: Sharpe, Omega, CVaR, Monte Carlo, Bloomberg

Systems: TCPDump, CUDA, OpenMP, Streamlit, S3

Credentials

• Stanford ML, DeepLearning.AI, Intel AI, Google CyberSec
• INSEAD Web3, Coursera Fin, Bloomberg Terminal
• TCPDump, Quantmod (R), SQL for Analysts

Recognition

• Top 1% – HL CS, ITGS
• Shazam M&A Analysis – Distinction
• Sigma.ai – AI instructor (HS level)
• MIT x OpenAI Hackathon Ambassador