Needl
An AI-assisted applicant tracking system: NLP résumé parsing and candidate scoring, built with a team of 18.
Built at GeekSI
- Spring Boot
- React
- TypeScript
- Flask
- spaCy
- AWS
Needl started as an R&D project at GeekSI. The idea was to help recruiters and candidates find each other faster by actually reading résumés instead of keyword-matching them. Since 2022 I've led it as solutions architect and tech lead, and we pitched it to partners in December 2025.
How it's built
- Application: Spring Boot with Hibernate over a MySQL schema, and a React + TypeScript front end.
- Intelligence: a Flask API wrapping an NLP résumé-parsing pipeline (spaCy) and ensemble scoring models (KNN, logistic regression, AdaBoost via scikit-learn).
- Infrastructure: AWS (VPC, ECS, ECR, Lambda, SQS, S3), all containerized with Docker.
What I learned
TODO(Dylan): the interesting part. What changed across the architectural rewrites, what you'd do differently, and what leading 18 people on a side-of-desk product taught you.