Sajan Poudel
I'm Sajan. I engineer AI solutions and am passionate about transforming industries through machine learning. My interests lie at the intersection of Reinforcement Learning, AI Agents, Quantitative Finance, and Space Exploration. I'm currently building production-level AI systems with multi-agent architectures and reinforcement learning optimization. Fueled by a deep curiosity for emerging technologies, I'm excited to leverage AI to drive forward-thinking solutions across diverse domains.
I've developed RocketSim, DreamWeaver, and various AI research projects, applying cutting-edge techniques to engineering, social platforms, and security systems. My work spans from intelligent CAD engines to fake news detection systems.
Projects
- RocketSim: AI-driven CAD engine with 6-agent architecture converting natural language to parametric 3D rockets with Monte Carlo simulation capabilities. [Web]
- DreamWeaver: AI-powered social platform leveraging LLM for dream semantic analysis and personalized content recommendations. [GitHub]
- CheatResume: AI-powered resume optimizer that tailors resumes to specific job descriptions for better matching. [Web]
- Asha: An AI health companion offering personalized health management through natural conversations. [GitHub]
- Cogniverse: A curated hub of AI tools and platforms for diverse applications. [Web]
Thoughts
- Multi-agent AI systems are the key to solving complex engineering problems autonomously
- Reinforcement learning will revolutionize how we optimize real-world systems and processes
- The fusion of AI agents with traditional CAD workflows will democratize engineering design
- Edge computing combined with blockchain creates unprecedented opportunities for secure distributed systems
- Fine-tuning open-source LLMs often outperforms proprietary models for specialized tasks
- The next breakthrough in AI will come from better agent coordination, not just bigger models
Publications
- Adaptive Proof-of-Stake DeFi Framework with Dynamic Multi-Asset Pools [Paper]
- Secure Scalable Provenance For Edge Computing Frameworks
- InDeKX: A High Efficiency Intrusion Detection and Knowledge Exchange Architecture for Secure Edge Computing Services [Paper]
Contact
Github |
Email |
LinkedIn |
Twitter