Inspiration
Interviews can be unpredictable, and candidates often struggle with nerves, uncertainty, and a lack of preparation. We wanted to create a smart, AI-driven solution that allows users to practice in a realistic, pressure-free environment. AceBot was built to simulate real interviews, generate dynamic questions, and provide insightful feedback—all within an intuitive, easy-to-use interface.
What it does
AceBot is an AI-powered interview coach that helps candidates prepare for both technical and non-technical roles. Here’s what it offers: ✔️ AI-Generated Questions – Generates tailored interview questions based on role and difficulty. ✔️ Real-Time Adaptation – Adjusts questions based on user responses, mimicking a real interviewer. ✔️ Multi-LLM Support – Uses LLaMA 3.1, Gemma 2, and Phi 3 for varied response styles. ✔️ Seamless User Experience – Built with Streamlit for an interactive, smooth interface.
How we built it
AceBot was developed using: 🔹 LangChain – For LLM-based question generation and response processing. 🔹 Ollama – To integrate multiple LLMs for dynamic conversation flows. 🔹 Streamlit – For a clean, interactive, real-time UI. 🔹 Python – To handle backend logic and AI processing.
Challenges we ran into
Ensuring diverse and relevant question generation across different domains. Fine-tuning LLM outputs to provide meaningful, structured feedback. Designing an intuitive UI that makes interview practice seamless and engaging.
Accomplishments that we're proud of
Successfully created a fully AI-driven interview simulation that adapts dynamically. Integrated multiple LLMs to provide diverse questioning strategies. Built a smooth, real-time interface with Streamlit for an optimal user experience.
What we learned
Effective LLM integration for structured question-answer flows. Balancing AI adaptability and realism in an interview setting. Optimizing AI-generated responses for better clarity and engagement.
What's next for Acebot
Expanding question categories to cover more specialized domains. ✔️ Enhancing response analysis with deeper AI-driven feedback. ✔️ Introducing voice-based interactions for a more immersive experience.
Built With
- ai-models:-ollama
- api
- gemma-2
- llama-3.1
- llm
- phi-3-frameworks-&-tools:-langchain
- streamlit-backend-&-logic:-python
Log in or sign up for Devpost to join the conversation.