Inspiration
Hiring is broken. Companies waste weeks screening candidates manually, and recruiters spend countless hours asking the same questions over and over. On the other side, candidates are often evaluated inconsistently or unfairly due to bias or fatigue. With the rise of powerful AI models, we saw an opportunity to build a scalable, consistent, and intelligent recruiting platform that automates the top-of-funnel experience — giving hiring managers real value and candidates a fairer shot.
What it does
- Conducts structured, intelligent interviews using conversational AI
- Performs real-time generative skill assessments and code tests
- Analyzes and scores candidate responses for soft and hard skills
- Produces a clear, consistent summary report for hiring managers
How we built it
Next.js + Tailwind for the frontend interface Redis - for storing the main information about the candidate and conversation history with AI model Node.js - for separate server that handle conversational based requests Websocket for message communication TTS ElevenLabs Whisper for audio transcription Postgresql
Challenges we ran into
Balancing speed, cost, and quality of AI API calls for real-time interaction Creating a natural experience that still maintains structure and professionalism
Accomplishments that we're proud of
- We've built fully candidate's part with real interactions with different API, Database and overall it almost ready-to-live prototype
- Built a fully functional conversational AI recruiter experience with structured logic
- Achieved realistic multi-turn interviews that adapt to candidate answers
- Integrated skill evaluation (both communication and coding) into a single flow
- Created a focused assistant that never loses sight of the hiring goal
- Developed a solid architecture that can scale with additional features
What we learned
- How to control large language models with precision using well-engineered prompts
- The importance of keeping sessions short, purposeful, and context-aware
- Real-world value of combining conversational UX with AI logic and scoring
What's next for SynthHire
- improve app for candidates
- create an admin panel for managers that will allow them to fully control the positions, candidates, and companies
- Introducing voice and video interfaces
- Adding interviewer scoring modes and company-custom prompts
- Expanding to role-specific flows: Product, Marketing, Sales, Design
- Provide vector search/vecror DB for providing better "memory" for AI
Built With
- nextjs
- node.js
- openai
- postgresql
- prisma
- react
- redis
- websocet
Log in or sign up for Devpost to join the conversation.