Computer science student building toward AI engineering from first principles. Three years of production web engineering gave me real systems instincts, now I'm applying that foundation to deep learning, model internals, and research implementation.

I care about understanding how things actually work, not just using APIs. That means reading source code, tracing gradient flow, understanding CUDA memory, and eventually contributing to the systems that make modern AI run.

Currently:

• Deep learning fundamentals via PyTorch (training loops, custom modules, optimizers)
• Reading AI research papers weekly — starting with the classics
• Working toward a pure NumPy ML library as a foundation capstone

On the horizon:

Transformer implementation from scratch · LoRA fine-tuning · Inference optimization · Agentic AI systems

• AI Engineering: PyTorch, deep learning architectures, training pipelines, and research paper implementation. Structured 26-month roadmap toward research-adjacent engineering.
• Systems depth: Leveraging C++ and Linux internals background to understand low-level ML: memory layout, CUDA execution model, efficient inference.
• Research literacy: Reading and engaging with AI papers weekly. Building the habit of connecting published ideas to working code.

Languages

Frontend

Backend

Databases

DevOps & Tools

Drop-in React code blocks with Shiki highlighting, Prettier formatting, and zero config. Dark and light themes, copy button, line numbers — all built in.

• VS Code-accurate syntax highlighting (TextMate grammars via Shiki)
• Automatic Prettier formatting on render
• Two components: <BxpCode> and <BxpCodeTabs>
• Zero-config defaults with full customization via props

Stack: TypeScript, React, Shiki, Prettier, Vite, VitePress

A premium collection of 4 elegant VS Code themes with unique identity, proper depth hierarchy, and signature accent colors. Includes dark themes (Obsidian, Sapphire, Onyx) and a light theme (Pearl).

• Optimized syntax highlighting for 15+ languages
• AMOLED-friendly true blacks option
• Eye-friendly light theme for daytime coding
• Distinctive bracket pair colorization

Stack: JavaScript, VS Code Extension API

Freelance Project Private Repository

Enterprise-style university portal built for a client, featuring hierarchical role-based access control.

• Multi-role architecture (Master Admin, Department Admin, Faculty, Student)
• JWT authentication with HTTP-only cookies + refresh token rotation
• Server-managed concurrent session limitation
• OTP-based password recovery with rate limiting
• Modular backend architecture (controllers + middleware)

Stack: JavaScript, MongoDB, Express, React, Node.js, Tailwind CSS

• Clean code and maintainable architecture
• Secure authentication and authorization systems
• Performance and scalability
• Developer experience and tooling
• Understanding systems from first principles