Hey there! I'm Aaditya Jindal—a tech enthusiast who codes like there's no tomorrow and debugs like there's definitely a tomorrow (usually at 3 AM). By day, I'm a computer science student with a passion for building things that actually work. By night... well, I'm still coding, just with more coffee ☕.
I specialize in turning "wouldn't it be cool if..." into production-ready applications. Whether it's fine-tuning LLMs on my laptop, building voice assistants for government schemes, or teaching robots to navigate mazes for accessibility, I'm all about leveraging technology to solve real-world problems—preferably before the hackathon deadline hits.
When I'm not pushing commits or wrestling with merge conflicts, you'll find me contributing to open-source projects, participating in hackathons, and continuously expanding my technical toolkit. I believe in writing clean code, building inclusive solutions, and never, ever pushing to production on a Friday.
🧩 ModelForge – No-code LLM Finetuning (repo)
A no-code toolkit to finetune LLMs on your local GPU—just upload data, pick a task, and deploy later. Built for fast iteration at hackathons and in early-stage prototyping, with automatic hardware detection and a guided React UI.
- 💡 Key features: task presets (classification, Q&A, chat), dataset upload with schema hints, experiment tracking, and one-click export of fine‑tuned weights.
- 🔐 Privacy by design: everything runs locally; no cloud dependency required.
- ⚙️ Hardware‑aware: automatic GPU/CPU fallback, mixed precision, and LoRA adapters for low‑VRAM machines.
- 🧩 Integrations: export an inference endpoint you can drop into your app.
- 🛠️ Tech stack: React, FastAPI, Hugging Face Transformers, PEFT, bitsandbytes, Docker (optional).
- 🎯 Why it stands out: true no‑code finetuning that actually works on laptops, enabling rapid model iteration without MLOps overhead.
💻 code_stream (repo)
A lightweight framework for real‑time code streaming and sync across Jupyter notebooks and services—ideal for live demos, classrooms, and pair‑programming.
- ⚡ Real‑time: stream cell edits and execution events as they happen.
- 🔌 Multi‑runtime bridge: TypeScript client SDK with a Python bridge for Jupyter.
- 🧱 Modular: minimal core with small, focused components you can extend.
- 🧪 Dev‑friendly: clean architecture, typed end‑to‑end, and easy to reason about.
- 🛠️ Tech stack: TypeScript/Node.js, Python, WebSockets, modern tooling.
- 🎯 Why it stands out: keeps scope tight—synchronization done right, without the bloat of full collab suites.
🎓 Learning Management System for NMTSA (repo)
Django‑based LMS for Neurologic Music Therapy Services of Arizona (NMTSA) featuring dual authentication, role‑based access, and accessible UI patterns tailored for neurodiverse learners.
- 🔐 Auth model: Auth0 for students/teachers; dedicated Django admin for administrators.
- 📚 Learning flows: courses, assignments, assessments, and submissions with progress tracking and gradebook views.
- 🧩 Roles & permissions: RBAC for students, teachers, and admins.
- 🎨 Accessibility: autism‑friendly UI with adjustable themes and font scaling.
- 🛠️ Tech stack: Django, HTML/CSS/JavaScript.
- 🎯 Why it stands out: designed with inclusive education in mind—simple, secure, and focused on essentials a small nonprofit actually needs.
🎨 DataCanvas – Paint Your Dataset! (repo)
An MS‑Paint‑style dataset builder to draw synthetic datasets up to 5 dimensions, persist them for later wrangling, and export to CSV for ML pipelines.
- ✏️ Visual authoring: sketch classes and clusters directly on a canvas.
- 🗃️ Persistence: save to MySQL for reproducible experiments.
- 📤 Export: one‑click CSV for training/validation workflows.
- 🛠️ Tech stack: Spring Boot (Java), HTML/CSS/JavaScript, MySQL.
- 🎯 Why it stands out: turns dataset prototyping into a fast, visual loop—no notebooks required.
🗣️ Saarthi – AI Voice Assistant for Government Schemes (repo)
A privacy‑first, voice‑forward assistant that helps citizens discover and understand Indian government schemes, with on‑device LLMs and secure identity.
- 🗣️ Multi‑modal I/O: converse in Hindi/English via text or voice.
- 🔐 Face authentication: protects access and personal context.
- 🧠 Orchestrated reasoning: LangGraph‑powered flows running local/private LLMs.
- 🖥️ Simple UI: Streamlit interface optimized for clarity and speed.
- 🛠️ Tech stack: Python, LangGraph, local LLMs, Streamlit.
- 🎯 Why it stands out: practical AI for public services—offline‑capable, private by default, and actually usable by non‑technical audiences.
🚗 ProjectSpyn – Autonomous Car Driving (repo)
A MATLAB robotics control system for LEGO Mindstorms EV3, built to improve mobility for specially‑abled individuals through personalized, autonomous maze navigation.
- 🔧 EV3 integration: motor/sensor control via MATLAB for deterministic behavior.
- 🧭 Autonomy: navigation routines for path planning and obstacle handling in a maze‑like city.
- ♿ Mission‑driven: inclusive design focused on assisted mobility.
- 🛠️ Tech stack: MATLAB, LEGO Mindstorms EV3.
- 🎯 Why it stands out: applies autonomy to a real accessibility challenge—not just a toy self‑driving demo.
- Portfolio & Projects: Devpost – Where I showcase my hackathon wins and projects
- Professional Network: LinkedIn – Let's connect!
- Tech Musings: X (Twitter) – Hot takes and tech discussions
- GitHub: RETR0-OS – You're already here!
Proud member of:
- 2025 Arizona Opportunity Hack – Building tech solutions for nonprofits and social good. Part of Team Coderz, where we developed an inclusive Learning Management System for NMTSA (Neurologic Music Therapy Services of Arizona).
Languages: Python, Java, C#, TypeScript/JavaScript, SQL, MATLAB, C
Frameworks & Tools: Django, Spring Boot, React, FastAPI, Node.js, LangGraph, Streamlit
AI/ML: Hugging Face Transformers, PyTorch, PEFT, LLM Fine-tuning, Computer Vision
Databases: MySQL, PostgreSQL
Other: Docker, Git, REST APIs, WebSockets, OAuth/Auth0, LEGO Mindstorms EV3
I'm always excited to collaborate on innovative projects, discuss the latest in AI/ML, or just geek out about technology. Whether you're a recruiter looking for someone who can ship production code and handle technical challenges, or a fellow developer wanting to collaborate—I'd love to connect!
Feel free to reach out via LinkedIn or check out my projects above. Let's build something amazing together! 🚀