MarkCoders · EDL Job Portal
AI Recruitment Platform
Production AI recruitment platform with resume intelligence, candidate matching, AI interviewing, and conversational AI — orchestrating OpenAI and Gemini behind a modular framework that cut inference cost ~50%.
The problem
Recruiters waste hours screening resumes, scheduling interviews, and chasing inconsistent candidate signals. Naive LLM integrations burn tokens on unprocessed documents, drive up latency, and are hard to extend when product requirements change.
How I solved it
I helped architect a production recruitment platform that treats AI as a pipeline, not a chat box. Resumes are crawled, preprocessed, and validated before inference. A modular multi-LLM router (OpenAI + Gemini) powers matching, interviewing, and conversational flows behind BullMQ jobs, Redis caching, and PostgreSQL persistence — cutting inference cost and response times by ~50% while making new AI features faster to ship.
Highlights
- Resume intelligence with preprocess-before-inference pipelines
- Multi-LLM orchestration across OpenAI and Gemini
- ~50% reduction in AI cost and latency
- Async workloads via BullMQ + Redis for reliable scale
Tech stack
- Next.js
- Multi-LLM
- BullMQ
- PostgreSQL
- Redis