Scribd
Upload
33 views6 pages

AI in Recruitment: Trends and Challenges

The document outlines the AALIM project, focusing on the development of an AI-driven recruitment platform for the education sector. It includes research summaries on AI applications in job matching, NLP for resume screening, and bias in AI assessments, alongside risk assessments and a detailed work division among team members. A timeline for project phases from requirement gathering to deployment is also provided.

Uploaded by

Rizwan Khan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
33 views6 pages

AI in Recruitment: Trends and Challenges

The document outlines the AALIM project, focusing on the development of an AI-driven recruitment platform for the education sector. It includes research summaries on AI applications in job matching, NLP for resume screening, and bias in AI assessments, alongside risk assessments and a detailed work division among team members. A timeline for project phases from requirement gathering to deployment is also provided.

Uploaded by

Rizwan Khan
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Assignment # 1

Project name: AALIM


Course: Software Project Management
Section: CS7A
Faculty: Prof. Shardha Nand
Members
Ali Talal (003)
Alishba Imran (117)
Mustafa Ilyas (267)
FYP Research Paper
1) Summary

This paper explores how artificial intelligence is increasingly applied in online matching systems,
documenting trends, key challenges and future directions. The authors analyses the growth of job-
resume matching models, emphasizing that while AI can enhance efficiency and fit in recruitment, it
also raises concerns around fairness, transparency and data quality. They highlight the importance of
structured profile data, feature selection, the impact of algorithmic bias, and the integration of ML
pipelines in real-world recruitment settings.
Reference:
Mert, M., Lehnert, K., & Güngör, T. (2022). A bibliometric perspective on AI research for job-résumé
matching. Frontiers in Artificial Intelligence, 5, 866441.

[Link]

2) Summary

This article investigates how natural language processing (NLP) techniques can support automated
résumé screening. The authors present methods for parsing raw teacher (or candidate) data,
transforming it into structured formats, extracting skills and experiences, and ranking suitability for
job roles. They demonstrate that combining lexical, syntactic and semantic features improves
matching accuracy and reduces manual screening effort. They discuss real-world constraints like
sparse data, varying document formats and evolving job descriptions.
Reference:
Patel, S., & Kumar, R. (2023). Resume screening with natural language processing (NLP). International
Journal for Research & Production, 6(6), 1234-1241. Retrieved from
[Link]

3) Summary

This research analyses membership (subscription) models in platform-based sharing-economy


contexts, focusing on when it’s optimal for a platform to offer premium membership to consumers,
providers, or both. Using a formal economic model they show: consumer-side memberships improve
social welfare more consistently, while provider-side or both-sides memberships can reduce welfare
and impose risks. Key leverage points: size of user base, supply-demand variability, marginal cost
dynamics, and the interplay of membership fees vs real-time service pricing.
Reference:
Zhang, H., Cavusoglu, H., & Raghunathan, S. (2025). Membership plans for sharing economy platforms.
SSRN.

[Link]

4) Summary

This paper presents a counterfactual-based framework to audit bias in AI-driven video interview
assessment systems. The authors show how multimodal inputs (video, audio, text) used in hiring‐
related AI tools can embed bias across gender, ethnicity, age. They propose using generative
adversarial networks (GANs) to create “counterfactual” versions of applicants by altering protected
attributes (e.g., gender) and observing changes in outcomes, thus quantifying unfairness in black-box
models. Their experiments show substantial disparities and recommend transparency tools for AI
hiring platforms.
Reference:
Mujtaba, D. F., & Mahapatra, N. R. (2025). Behind the screens: Uncovering bias in AI-driven video
interview assessments using counterfactuals . arXiv.

[Link]
FYP Ideas
• 1. A Novel Approach for Learning How to Automatically Match Job Offers and Candidate
Profiles — Jorge Martinez-Gil, Alejandra Lorena Paoletti, Mario Pichler (2016)

Link: [Link]

• 2. Fairness in AI-Driven Recruitment: Challenges, Metrics, Methods, and Future Directions —


Dena F. Mujtaba & Nihar R. Mahapatra (2024)

Link: [Link]

• 3. The Impact of Artificial Intelligence (AI) on Recruitment Process — Umaima El Ouakili


(2025)

Link: [Link]

Reactive Risks
Risk Description Mitigation Approach

Data bias & The matching/assessment algorithms may embed Use diverse training data, bias‐audits,
fairness biases (Gender, Areas, Cities, Provinces) and periodic reviews, transparency in algorithm
reduce fairness. (See fairness paper above) decisions.

Privacy & legal Collecting video interviews, quiz data, NLP on CVs Define clear privacy policies, anonymize
compliance means handling personal data—risk of non‐ data, follow local/international data
compliance with regulations. protection laws, get consent, security
safeguards.

Technical Modules (backend, frontend, AI) may not Regular integration checkpoints, modular
integration integrate smoothly (due to API version architecture, early prototyping of interfaces,
delays mismatches, third-party services). buffer time in timeline.

User adoption Teachers or institutes might resist using a new AI Conduct user research, provide training,
resistance system (fear of automation, trust issues). demonstrate value clearly, include human–
in-loop options, gather feedback early.

Model The AI matching, assessment, interview analysis Use pilot runs, refine models iteratively,
accuracy & might give false positives/negatives, undermining include fallback or manual review, monitor
reliability trust. performance metrics continuously.

Productive Risks
Risk Potential Opportunity How to Leverage

Rapid tech If you adapt quickly, you can become early- Build modular AI architecture, keep
changes mover and differentiate your platform. research/design flexible, monitor emerging
(AI/NLP) models (e.g., zero-shot matching).

High user If you exceed user expectations (smooth UX, Priorities UX work (Member1), gather user
expectations meaningful insights), you gain strong word- feedback early, iterate UI/UX aggressively.
of-mouth and retention.

Competitive Many generic portals exist, but niche Emphasize niche angle, build unique features
pressure (teachers + institutes) with AI is less common (interactive CV builder, video interview, auto
=> big market for you. quiz) and market them as unique selling
points.

Data As you collect feedback/performance data you Build in logging/analytics from day one,
accumulation can refine models, create proprietary insights, create feedback loops (Member3) and
over time improve matching accuracy. leverage metrics to enhance system.
Research Questions

1. How can machine learning be used to enhance the accuracy and efficiency of teacher–job matching?

2. How effectively can AI transform unstructured teacher data into professional resumes?

3. How can AI auto-grades systems contribute to fair and real-time teacher assessments?

4. What are the usability and performance benefits of a dual-dashboard design for teachers and institutes?

6. To what extent can the AALIM platform centralize the teacher recruitment process and eliminate external
dependencies?

Research Objectives

1. To develop an AI-driven recruitment platform tailored for the education sector that connects teachers with
institutions effectively.

2. To utilize natural language processing (NLP) for automatic and structured CV/resume generation from unformatted
teacher data.

3. To integrate AI-based Resume Generator or Enhancing. And using AI and ML to suggest best suitable Candidate
for Job.

4. To build a dual-dashboard system (Teacher and Institute) ensuring a unified, user-friendly recruitment
environment.

5. To evaluate the overall performance, fairness, and efficiency of the proposed AALIM system compared with other
competitive platforms.

Work Division
Member 1: Frontend Engineering and Development

1) Design and develop responsive dashboards (Teacher + Institute).

2) Implement interactive CV builder interface.

3) Integrate real-time video interview UI (WebRTC/Zoom API).

4) Ensure usability, accessibility, and smooth user experience.

Member 2: Backend Development & Database Management

1) Build secure REST APIs for authentication, job postings, and user

management.

2) Design and optimize relational database (PostgreSQL/MySQL).

3) Handle data flow between dashboards and core modules.

4) Implement role-based access control (teachers vs. institutes).

Member 3: Artificial Intelligence & Machine Learning Integration

1) Develop AI-based teacher–institute matching using ML algorithms.

2) Implement quiz auto-grading and intelligent feedback.

3) Build NLP-powered CV analysis for skill extraction.

4) Research & integrate pretrained models (e.g., BERT, scikit-learn classifiers).


Member 4: Testing & Documentation and UI/UX Designer

1) Conduct unit testing, integration testing, and system-level testing.

2) Manage bug tracking and ensure system reliability.

3) Prepare user manuals and technical documentation.

4) Oversee deployment pipeline and final project report.

Timeline
Month Phase Key Activities

Month 1 Requirement gathering & Stakeholder interviews (teachers & institutes), define personas,
analysis compile detailed feature list (matching, CV builder, assessment,
dashboards)

Month 2 Feasibility study & technology Evaluate ML/NLP models, video-interview tech (speech-to-text,
selection sentiment), quiz generation tools; pick tech stack

Month 3 System design – architecture & Design system architecture (dashboards, APIs, data flows), design
database relational DB schema for users, jobs, assessments, interviews

Month 4 UI/UX design – wireframes & Create teacher dashboard + institute dashboard wireframes, design
mockups interactive CV builder UI, get user feedback

Month 5 Backend development – core Build user authentication, role-based access, job posting logic,
modules matching engine stub, data API endpoints

Month 6 Frontend development – Develop dashboards for both user types, implement CV builder UI,
candidate & recruiter interfaces integrate some APIs

Month 7 Integration of backend + Connect UI with backend services, integrate matching engine, test
frontend flows for teacher/institute, fix integration bugs

Month 8 Online assessment module Implement quiz generation, auto-grading engine, integrate with
development teacher dashboard, collect initial data

Month 9 Interview scheduling & video Build scheduling module, integrate video interview UI (speech-to-
module development text + sentiment analysis), link results to matching engine

Month Testing phase – system testing Conduct full system testing (unit, integration, usability), do pilot with
10 & bug‐fixing sample users, track bugs, refine features

Month Deployment & pilot run Deploy system for pilot use by few institutes & teachers, gather user
11 feedback, monitor key metrics (matching accuracy, user satisfaction)

Month Final review & documentation Final project report, user manuals, technical documentation, refine
12 based on pilot feedback, prepare for launch
Chart

-------------------------------------END---------------------------------

Common questions

Powered by AI

NLP enhances résumé screening efficiency by parsing and transforming raw data into structured formats, extracting skills and experiences, and improving ranking accuracy. It faces constraints such as sparse data, varying document formats, and the need for adaptability to evolving job descriptions . By integrating lexical, syntactic, and semantic features, NLP can significantly reduce manual screening efforts .

The main challenges in using AI for job-resume matching include fairness, transparency, data quality, and algorithmic bias. These can be mitigated by ensuring that structured profile data is used, selecting relevant features carefully, and integrating machine learning pipelines that address real-world constraints. Fairness can particularly be improved through diverse training data and bias audits . Transparency can be enhanced by providing clear explanations for the AI's decisions and using transparency tools for AI hiring platforms .

Consumer-side memberships are shown to improve social welfare more consistently than provider-side or both-side memberships. This is because consumer-side memberships tend to leverage a larger user base and exhibit less variability in supply-demand dynamics. This helps maintain marginal cost balances more effectively, unlike provider-side memberships which can reduce welfare and pose additional risks .

Counterfactual analyses can audit AI interview systems for bias by creating hypothetical scenarios where protected attributes, such as gender or ethnicity, are altered. By observing changes in the system's evaluations under these altered conditions, analysts can quantify unfairness and identify specific sources of bias. This method provides a systematic approach to understanding and reducing bias in AI systems .

Rapid technological changes pose risks like technical integration delays and user adoption resistance due to fear of automation. However, they also present opportunities such as becoming an early mover with differentiated platforms by swiftly adapting to these changes. Recruitment platforms can leverage modular AI architecture and flexible design to capture emerging models and exceed user expectations through superior UX and insights .

A dual-dashboard system is designed to offer a unified and user-friendly recruitment environment for both teachers and institutions. This design aims to enhance usability and performance by ensuring that both parties have tailored interfaces that support their specific needs and contribute to smoother interactions and streamlined processes .

Recruitment platforms can ensure compliance by defining clear privacy policies, anonymizing collected data, following local and international data protection laws, obtaining consent from individuals, and implementing security safeguards. This approach helps manage the risks associated with collecting video, quiz data, and performing NLP on resumes .

Modular AI architecture allows for the flexible integration of new models and technologies as they emerge, thereby enhancing adaptability. By breaking down the system into interchangeable modules, the platform can easily incorporate innovations in AI and NLP, facilitating rapid adjustments to meet evolving market demands and user expectations, ultimately maintaining a competitive edge .

Bias in AI-driven video interview assessments can be detected using counterfactual-based frameworks, which analyze multimodal inputs like video, audio, and text. Generative adversarial networks (GANs) can create counterfactual versions of applicants by altering protected attributes such as gender, to observe changes in assessment outcomes. To mitigate bias, transparency tools that elucidate the model's decision processes and regular bias audits are recommended .

Feature selection plays a crucial role in mitigating algorithmic bias by ensuring that the model only uses relevant and unbiased features for decision-making. By selecting appropriate features, models can avoid perpetuating biases that might originate from irrelevant or historically biased data. This careful selection can lead to more effective and equitable recruitment processes .

You might also like