InterHacker - MCP

Inspiration

We wanted to reimagine technical interview prep, making it adaptive, voice-interactive, and context-aware. Instead of static question banks, we envisioned an AI interviewer that truly knows you—your code, resume, and target role, helping you grow like a real mentor would.

What it does

InterHacker is an MCP-based mock interview assistant that dynamically tailors interviews to each user. It pulls data from your GitHub repos, resume, and job postings to generate intelligent technical and behavioral questions. It evaluates answers for correctness and communication, and even spins up live coding challenges on Vultr.

How we built it

We built a FastAPI-based MCP server with: LangGraph ReAct Agent (Gemini) for dynamic question generation and adaptive follow-ups. MongoDB Atlas for session persistence and analytics. Vultr for hosting and sandboxed coding challenge environments. Snowflake (optional) for long-term performance analytics and skill insights. Everything communicates through an MCP WebSocket endpoint, allowing external tools or clients to plug into the interview engine seamlessly.

Challenges we ran into

Designing a scalable, modular MCP interface that could handle live audio input and AI-driven follow-ups. Context synthesis—balancing detail from resumes, GitHub repos, and job descriptions without overwhelming the model. Integrating Vultr compute securely for ephemeral coding tests.

Accomplishments that we're proud of

Built a fully functional MCP-compliant server integrating AI reasoning, voice I/O, and contextual retrieval. Created an adaptive feedback engine that analyzes both content and communication style. Achieved seamless orchestration between MongoDB, LangGraph, ElevenLabs, and Vultr. Developed a scalable foundation for personalized AI coaching.

What we learned

How to design context-rich AI experiences that blend multiple modalities—voice, text, and code. The importance of structured MCP protocols in building modular, pluggable AI ecosystems. Techniques for balancing generative creativity with deterministic evaluation and scoring. Deploying multi-component architectures on Vultr efficiently.

What's next for InterHacker

Real-time coding evaluation with AI-assisted hints and debugging. Deep analytics dashboards for tracking skill growth. Multi-language interview modes (Python, C++, Java). Integration with ATS/job platforms for custom prep paths. Fine-tuned voice personalities for tailored interviewer styles.