I don't just push boundaries; I gave them an eviction notice. Welcome to the lab.
Look, I don't just "code"—I teach rocks how to think and get them to thank me for it. I’m a founder-engineer who treats production like a jazz solo: calculated chaos with a lot of soul. I build high-ROI AI products because "impossible" is just a dare I haven't taken yet. My goal? Automate the mundane, scale the genius, and build the kind of tech that makes Skynet look like a pocket calculator. If you're looking for safe, color-by-numbers engineering, go hire a spreadsheet. I'm here to ship fast, break the status quo, and turn ungodly amounts of caffeine into pure, monetizable magic.
Right now, the project getting most of my obsession is OpenJCK (formerly AgentTrace): the observability + reliability runtime for autonomous AI agent systems — basically Cloudflare for AI agents.
If NVIDIA OpenShell answers "what can an agent do?", OpenJCK answers "what did it do, why did it fail, and how do we keep fleets alive 24/7?" It’s designed for the new always-on world of Claude-based autonomous agents, where reliability is not a feature — it’s survival.
- ✅ v0.1 shipped:
@tracedecorator, static HTML viewer, npm package - 🔨 v0.2 in progress: live dashboard, Failure Intelligence, Express + SQLite backend, SSE
- 🧭 Next: the five-phase roadmap to production-grade agent infrastructure
- 🔭 Currently Architecting: Axiom-One (Research-Grade Reasoning System—because chatbots are boring) & Agent-X (Vision-Language Browser Agent—yes, it can see you).
- 🌱 Downloading to Brain: Agentic AI, LangGraph, Fine-tuning... basically learning how to make the machines do my chores.
- 🤝 Looking for Co-Conspirators: To build innovative AI projects or anything that sounds vaguely impossible.
- 💬 Provoke Me About: System architecture, the singularity, or why AI ethics is the only conversation worth having at parties.
- ⚡ Weird Flex: I've fine-tuned 8 AI models, swept 3 hackathons, and type at 120 WPM. My blood stream is 90% Arabica.
|
Oct 2024 - Dec 2024 AI agent that automates browser tasks using vision-language models. Interacts with websites from plain English instructions, leveraging Python, FastAPI, Playwright, React, and Gemini 2.5. Open-source, safety-first, and fully browser-native for hands-free web automation.
|
Jan 2025 - Present The observability and reliability runtime for autonomous AI agent systems. Add
|
|
Jun 2024 - Sep 2024 Comprehensive RAG course covering theory, production-ready code, and 15 modules—reranking, chunking, embeddings, vector DBs, hybrid search, query optimization, agentic workflows, and deployment. Suited for AI/ML developers seeking advanced RAG knowledge.
|
Aug 2024 - Dec 2024 A Research-Grade Reasoning System (Not a Chatbot). A thinking-first research agent that decomposes problems, runs multi-step reasoning loops, validates its own outputs, and exposes the entire thought pipeline. Optimizes for correctness under pressure.
|
- 🎓 Building toward Computer Use with Anthropic - Anthropic (Mar 2025)
- 🏢 Software Engineering Job Simulations - EA, Walmart, Goldman Sachs (Forage, 2024)
- 🤖 AI Agents in LangGraph - DeepLearning.AI (2024)
- 🚀 Finetuning Large Language Models - DeepLearning.AI (2024)
- 🎨 Front End Development Libraries - freeCodeCamp (Feb 2025)
- 👁️ Prompt Engineering for Vision Models - DeepLearning.AI (Dec 2024)
- 🧠 Machine Learning with Python - Cognitive Class (Dec 2024)
- 🔗 Blockchain Masterclass - CFTE
- 💼 LinkedIn: Ravani Roshan (My "Professional" Mask)
- 🌐 Portfolio: ravani-roshan-singh.vercel.app/certificates (The "I'm Actually Good" Proof)
- 📧 Email: ravaniroshansingh@gmail.com (Send memes or job offers. Preferably both.)
- 💻 HackerRank: Profile
"I don't just solve problems. I create solutions that make the original problem look like it was asking for it." 💪
Fair warning: I may accidentally automate your job while we're talking. 🌟