AI-Based Resume Optimizer Platform using
Spring Boot, Flask, and React
S. Nirmal,
Department of Computer Science and Engineering
Anand Institute of Higher Technology, Kazhipattur, Chennai
nirmal7048@[Link]
Abstract— This paper presents an AI-Based Resume
Optimizer Platform built on a microservice II SYSTEM ARCHITECTURE
architecture. The system integrates a Spring Boot The platform’s architecture is designed for scalability,
backend for business logic and data persistence, a modularity, and ease of integration, following microservice
React frontend for the user interface, and a Flask- principles widely adopted in enterprise-grade AI systems. It
based AI bridge. The backend ingests resumes using enables distributed processing of resumes, independent
Apache Tika and orchestrates analysis. The Flask scaling of services, and seamless orchestration across
bridge leverages NLP libraries (e.g., spaCy) to multiple layers.
perform ATS scoring, job-match evaluation,
grammar correction, and keyword-gap detection. The system adopts a three-tier microservice architecture
The React frontend provides a secure (JWT) comprising three cooperating components:
dashboard for users to upload documents, review
analysis history, and download enhanced reports. a Spring Boot backend that manages
This platform provides actionable, AI-driven authentication, resume storage, and workflow
guidance to streamline resume preparation. orchestration,
a Flask-based AI Bridge that encapsulates natural-
Keywords— Resume Optimization, AI-Driven language pipelines for resume analysis,
Analysis, Spring Boot, Flask, React, ATS Scoring, and a React-based frontend that delivers
NLP. dashboards and interactive user experiences.
I. INTRODUCTION Each module communicates through well-defined REST
The Resume Optimizer Platform tackles the persistent challenge interfaces, enabling asynchronous processing and
job seekers face in crafting resumes that align with component independence. This design ensures that each
automated Applicant Tracking Systems and nuanced service can evolve, scale, or be redeployed without
recruiter expectations, a process often derailed by impacting the others. For instance, the Spring Boot backend
inconsistent formatting, overlooked keywords, and can operate behind a load balancer, the AI Bridge can be
limited feedback loops. Motivated by advances in natural- containerized and replicated with GPU support for heavy
language processing and the scale at which AI can inference tasks, and the React frontend can be served via a
evaluate textual quality, the system integrates an AI- CDN for rapid client delivery.
driven analysis service with robust storage and an
interactive web interface to deliver actionable insights— DAIGRAM 1
spanning ATS scores, job-match alignment, grammar
Resume Optimizer Platform
diagnostics, and keyword recommendations—
immediately after upload. Within this scope, the platform
aims to streamline resume preparation end-to-end:
securely ingesting documents, extracting and persisting
structured data, invoking AI meta-analysis through the
Flask AI Bridge, archiving historical results in the Spring
Boot backend, and presenting users with intuitive controls
in the React frontend so they can repeatedly refine and
download enhanced resumes.
To further ensure practical scalability, the system adopts a
modular microservice architecture that separates concerns
between analysis, orchestration, and user interaction layers.
This design not only supports extensibility for future AI
model integration but also ensures reliability and faster
response times under concurrent user workloads.
A. Layered Architectural Design III LITERATURE SURVEY
The architecture is logically divided into three layers, as In recent years, artificial intelligence (AI) and natural
shown in Fig. 1, to promote separation of concerns and language processing (NLP) have played a transformative
simplify maintenance. role in recruitment, particularly in automating resume
evaluation, candidate–job matching, and skill analysis.
1) Presentation Layer (Frontend)
Numerous research works have proposed intelligent
The React + Vite frontend acts as the presentation layer, systems that enhance resume quality and streamline hiring
offering a modern user interface that supports resume through data-driven insights.
uploads, interactive dashboards, and result visualizations.
Through RESTful API calls using Axios, it securely The study by [1] V. Keerthiga et al. (2024) proposed a
communicates with the backend, handling token-based Career Guidance Web Application that assists students and
authentication and rendering analytics in real time. job seekers in building resumes, receiving real-time AI-
Animations, hooks, and responsive design elements based feedback, and obtaining career recommendations
enhance usability and provide real-time feedback on through a Random Forest Classifier (RFC) model. While
upload and analysis status. the system achieved 91% accuracy, its scalability and
integration of advanced NLP feedback mechanisms
2) Application Layer (Backend and AI Bridge) remained limited.
The Spring Boot backend serves as the core orchestration
engine, handling authentication (via JWT), API routing, and [2] Bevara et al. (2025) introduced Resume2Vec, a deep-
communication between the user and AI services. It also learning framework leveraging word embeddings to
manages the scheduling of analysis tasks, resume represent resumes semantically for improved job–candidate
persistence, and history logging. matching. This approach enhances contextual understanding
but lacks a comprehensive front-end interface for interactive
The Flask AI Bridge represents the intelligence layer
resume improvement and real-time grammar diagnostics.
within the application architecture. It encapsulates
multiple AI modules for: [3] Changmao Li et al. (2024) developed ConFit, a resume-
ATS Scoring: Evaluating resume structure and job matching framework based on contrastive learning and
keyword density, data augmentation. Although it achieves higher precision in
Grammar Feedback: Detecting and correcting candidate recommendation, the system primarily focuses on
grammatical inconsistencies using NLP, backend AI models without providing end-user usability or
feedback mechanisms.
Keyword Extraction and Job Matching:
Identifying missing skills and ranking job-fit scores. [4] Junshu Jiang et al. (2020) presented a Feature-Fusion
This layer is designed to scale horizontally and supports Model that integrates explicit and implicit representations
experimentation with different NLP models or ranking for person–job fit prediction. While the model performs well
algorithms without altering backend or frontend in matching accuracy, it lacks an integrated platform
configurations. combining resume analytics, grammar checks, and job-fit
scoring.
3) Data Layer (Database and Storage)
The MySQL database serves as the persistent data layer, [5] Singh et al. (2023) designed an AI Resume Shortlisting
managing users, resumes, and analysis history through System using Support Vector Machines (SVM) for ATS
relational tables. It ensures data integrity through foreign ranking. Their system provides foundational keyword-based
key constraints and transactional consistency via Spring ranking but lacks multi-dimensional features such as real-
DataJPA. time grammar feedback, historical analysis tracking, or
Additionally, the system provides abstraction for local modular architecture.
and cloud storage, allowing resume PDFs, JSON reports,
and historical logs to be securely maintained under user-
specific directories or cloud object stores (e.g., Google
Cloud Storage or AWS S3).
B. Extensibility and Research Scalability
This modular separation enables future system expansion,
such as integrating AI services for interview question
generation, resume summarization, or semantic job-
matching using transformers. The consistent schema
contracts between modules guarantee interoperability
while supporting new research experiments.
Overall, this architecture establishes a cloud-deployable,
research-friendly, and AI-driven resume optimization
ecosystem.
TABLE 1 TABLE 2
Related Research Works System Component Breakdown
Communica
Module Tech
Pa per Methodology / Key Focus / Access Key Features tion
Stack
& Au Model Findings Refere Protocol
tho r nce JWT authentication,
Link Spring resume storage, text
Backend
Boot extraction (Apache REST/
(Java) Tika), and AI HTTP
V. Random Forest AI-based resume IEEE orchestration
Keert higa et Classifier builder, Xplore DOI:
al. (RFC) for chatbot for 10.1109 NLPanalysis (spaCy),
ATS
resume career /GCAT AI Flask
scoring, grammar REST/
screening and guidance, 91 62922.2 Bridge (Python)
evaluation, and skill HTTP
recommendati % accuracy; 024.109
recommendations
on. limited NLP 23842
grammar Interactive dashboards,
support. resume
React
Frontend upload/analysis/do
R. V. K. BERT-based Improves semanticMDPI (TypeScr Axios/
wnload, real-time
Bevar a et al. word relevance in Electro nics ipt) HTTP
feedback
embeddings ATS (Free visualization
and vector pipelines but Access) Stores user data,
similarity lacks user- Database resumes, analysis
for job- interactive MySQL JDBC
history, and job-
candidate front matching metadata
matching. end.
Chan gmao Li Contrastive Enhances arXiv:
et al. learning and matching 2401.16 4.2 Backend Implementation
data accuracy; 349 (PDF) The Spring Boot backend serves as the core orchestration
augmentatio backend-only layer, managing authentication, file handling, and
n for robust design communication with the AI Bridge service. The
person–job without UI or architecture of this module is depicted in Figure 1.
fit feedback.
The primary components include:
models.
Junsh u Jiang Feature-fusion Accurate person– IEEE ResumeController: Handles /upload, /analyze, and
et al. neural job fit Access /download endpoints.
network prediction; (Free AnalysisController: Routes AI analysis requests
combining lacks PDF) and persists JSON results.
explicit + grammar/key FileStorageService: Manages local or cloud-based
implicit wor d storage of uploaded resumes.
features. feedback
SecurityConfig: Implements JWT-based
modules.
authentication using Spring Security.
Key Code Snippet — Resume Analysis Logic
II. IMPLEMENTATION
4.3 AI Bridge Implementation
4.1 System Implementation
The proposed Resume Optimizer Platform was developed The AI Bridge, implemented in Flask, acts as an
using a modular microservice-based architecture independent inference gateway. It exposes
consisting of four core components, as shown in Table 1. multiple REST endpoints that perform ATS
Each component communicates via REST APIs to scoring, grammar evaluation, and keyword-based
maintain separation of concerns, scalability, and ease of skill recommendations. Figure 2 shows its
deployment.
pipeline workflow.
TABLE 3
Key Endpoints
Endpoint Input Output
{ats_score,
resume_text, grammar_sugges
/analyz
job_descript tions,
e/meta
ion missing_keywor
ds}
{recommended_s
/recomm
resume_text, kills,
end/ski
target_job improvement_ti
lls
ps}
The service integrates spaCy for NLP processing,
tokenization, and grammar correction, ensuring
lightweight yet accurate linguistic feedback.
Responses are standardized into JSON schemas,
allowing seamless persistence by the backend.
4.4 Frontend Implementation
The React-based frontend provides a user-friendly
interface that enables resume uploads, real-time
analysis, and historical result visualization. The
frontend design follows a component-driven
architecture , with hooks for state management and
Framer Motion for animations.
Core Components:
[Link]: Displays uploaded resumes
and their scores.
[Link]: Triggers AI
analysis and visualizes the results
dynamically.
API Client: Implements Axios with JWT
token injection for secure communication.
Frontend Flow (Mermaid Diagram)