Siddharth Chaudhary

I am an independent researcher in my final year at St Paul’s School, London. My research centers on foundational sequence learning and adaptive architectures, including transformers, state-space models, and recurrent networks, with an emphasis on modularity, memory formation, and plasticity mechanisms that support efficient long-context reasoning and rapid adaptation.

I have currently authored two papers, and am actively working on more:

Hydra: A Modular Architecture for Efficient Long-Context Reasoning — NeurIPS 2025 Workshop on Efficient Reasoning.
Enabling Robust In-Context Memory and Rapid Task Adaptation in Transformers with Hebbian and Gradient-Based Plasticity — Under Review at TMLR 2025

My previous experience involves working at startups involved in applied ML, including Cactus (YC S25), and CorpusAnalytix (Wharton Cypher Accelerator)

This website is currently a work in progress, but in the meantime feel free to browse some of my solutions to Project Euler problems.

Resume

You can find my resume in pdf format here, or view below.

Siddharth Chaudhary

chaudhs@stpaulsschool.org.uk | linkedin.com/in/sidcraftscode | github.com/sidcraftscode

Education

St Paul's School, London, UK
General Certificate of Education, 2021 – 2026
- A-Levels: Mathematics, Further Mathematics, Physics, Computer Science, all predicted A*
- Extended Project Qualification: Designing and developing an iOS app to track workouts, achieved A*
- GCSEs: 11 in total including Mathematics and English Language, all grade 9 (equivalent to A*).
- Olympiads: UK Olympiad in Artificial Intelligence 2025; achieved Gold (92/100) and qualified for national team selection interviews.

Experience

Independent Researcher, St Paul's School, London, UK
Sept. 2025 – Present
- Focus: Foundational sequence learning and adaptive architectures, including transformers, state-space models, and recurrent networks.
- Conducting independent research on modularity, memory formation, and plasticity mechanisms for efficient long-context reasoning and rapid adaptation.
- Authored two papers:
  Hydra: A Modular Architecture for Efficient Long-Context Reasoning – NeurIPS 2025 Workshop on Efficient Reasoning.
  Enabling Robust In-Context Memory and Rapid Task Adaptation in Transformers with Hebbian and Gradient-Based Plasticity – Under review at TMLR, 2025.
- Implemented large-scale experiments using self-managed GPU infrastructure.
- Released reproducible open-source implementations accompanying both papers.
Co-founder, Eleven Plus Vocab, London, UK
Aug. 2023 – Present
- Co-founded and currently operating an edtech venture providing digital products for UK 11+ entrance exam preparation.
- Built and maintained an e-commerce platform using Next.js integrated with the Shopify API, generating thousands in annual revenue.
- Developed companion iOS and Android apps for practice and assessment, reaching several thousand downloads.
Software Engineering Intern, Cactus Compute (YC S25), San Francisco, CA
July 2025 – Aug. 2025
- Implemented end-to-end tool-calling capabilities in the core Cactus framework, from API design to integration and testing.
- Contributed to the open-source library and React Native demo app showcasing new inference and framework features.
- Led testing and provided structured feedback on framework releases to improve reliability and developer experience.
Software Engineering Intern, CorpusAnalytiX (Wharton Cypher Accelerator), Philadelphia, PA
Mar. 2025 – June 2025
- Contributed to a healthcare data ecosystem designed to democratize research access and accelerate data-driven R&D.
- Developed intuitive user interfaces in TypeScript and Next.js for dataset exploration and request management.
- Fine-tuned language models to generate synthetic dataset “heads,” allowing secure, privacy-preserving previews of sensitive medical data.
Full-Stack Developer, Lgcy, San Francisco, CA
Aug. 2024 – Mar. 2025
- Designed and implemented the app’s user interface and visual system with a focus on accessibility and cohesive experience.
- Developed and integrated GPT-based conversational agents using fine-tuned models and the OpenAI API.
- Collaborated on product architecture and conducted performance optimization and code reviews to improve reliability.

Projects

MusclePeak | Swift, SwiftUI, CoreData, Xcode, iOS
Sept. 2024 – May 2025
- Developed and released a fitness app for iOS that allows users to log workouts, build multi-week training plans, and track progress efficiently.
- Designed a minimalist, intuitive UI with offline data storage, automatic rest timers, and advanced set-tracking features.
- Published to the App Store: https://apps.apple.com/gb/app/musclepeak/id6741804742

Technical Skills

Languages: Python, JavaScript/TypeScript, Swift, C/C++, SQL
ML: PyTorch, JAX, Matplotlib, NumPy, pandas, scikit-learn
Frameworks: React, Next.js, Node.js, React Native, SwiftUI, UIKit
Tools: Git, Docker, Vercel, Xcode, Figma, MongoDB