Applying prompt engineering, fine-tuning, and experimentation to refine model performance, accuracy, and reliability, ultimately increasing the efficacy and applicability of AI across CrowdDoing’s initiatives.

» Collaborated on developing multi-agent applications leveraging open-source frameworks (e.g. Autogen, AgentVerse, MetaGPT) to facilitate task-oriented and simulation-oriented conversational interactions within Large Language Model (LLM) based autonomous agents.

» Optimized performance of open-source large language models (LLMs) using techniques like prompt engineering and fine-tuning, leading to improved accuracy, reduced hallucination, and a 50% task completion rate for autonomous agents.

» Leveraged open-source LLMs like LLaMa and Mistral on LINUX environment to empower autonomous agents to complete given tasks.

» Designed and implemented an environment for multi-agent systems to evaluate agent performance by simulating human behavior through experience storage, memory synthesis, and dynamic retrieval for behavioral planning.

» Assisted with web development services and website management tasks, gaining valuable experience in real-world web project workflows.

» UtilizedDjangoandJavaScripttodevelopwebAPIstocreateinteractivewebexperiences.