International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056
Volume:12Issue:12|Dec2025 www.irjet.net p-ISSN:2395-0072
KaamKhojAI: AI-Driven Employment Accessibility for Low-Literate Job Seekers
Manthan Tighare 1 , Yachin Verma2 , Pratham Potdar3 , Rishabh Prajapati4 , Aashish Choudhari5 , Prof. Swati Dronkar 6
6Professor, CSE, Priyadarshini College of Engineering Nagpur, Maharashtra, India 12345UG Student, CSE, Priyadarshini College of Engineering Nagpur, Maharashtra, India
Abstract - India’s working population comprises individuals from a wide range of professional and skill-based categories, includingunskilledlaborers,serviceworkers,technicians,and educatedprofessionals.Despitetechnologicaladvancements, many job platformsstill lack accessibility featuresto support diverse user groups who struggle with language barriers, digitalliteracy,andcomplexregistrationprocesses. Artificial Intelligence (AI), Natural Language Processing (NLP), and conversationaltechnologiesaretransformingtherecruitment sector by enabling AI-driven assistance, automated interactions, and personalized job recommendations. This reviewpaperanalyzestheevolutionofAI-assistedjobportals designed to support all categories of workers in India, examining their architectures, features, and usability components that simplify employment access. The study identifies key limitations such as insufficient multilingual support, weak contextual job matching, and limited regional employment integration, while emphasizing the need for inclusive and scalable AI solutions for employment accessibility.
Key Words: AI Assistant, Job Recommendation, Speech Recognition, MERN Stack, Workforce Inclusion,EmploymentAccessibility
1. INTRODUCTION
Nagpur,beingarapidlydevelopingmetrocityandamajor logistics hub of Maharashtra, hosts a diverse workforce ranging from daily wage laborers and manufacturing workerstoretailstaff,technicians,andserviceprofessionals. However,mostpopularjobportalssuchasLinkedIn,Indeed, andNaukriremainorientedtowardurbanwhite-collarjob seekers, limiting their usage among Nagpur’s semi-skilled and informal workers who may not be comfortable with digital interfaces. An AI-based job portal specifically designed for the Nagpur region can bridge this gap by offering regional language guidance, simplified registration, and AI-assisted job searching, making employment discovery effortless for all categories of job seekers.ThisreviewconsolidatespreviousresearchonAIenabled employment systems and analyzes how these technologiescanbesuccessfullyadaptedforNagpur'slocal employmentecosystem.
2. Methodology of Review
This review is based on academic research published between2014and2025,sourcedfromGoogleScholar,IEEE Xplore,ResearchGateandManymore.Studieswereselected according to three major criteria: they must address technologies that improve job searching, include AI-based user interaction features, and discuss accessibility for workers with limited digital proficiency. A broad range of research papers were reviewed to understand the development of AI-assisted job platforms. Among them, several studies were found highly relevant due to their emphasis on conversational AI support, intelligent recommendation systems, and solutions tailored for users with limited technical skills. These selected works were examined to compare emerging technological trends, architectural approaches, user experience improvements, and the existing limitations affecting the adoption of AIdrivenemploymentportals.
3. Literature Review
3.1JobPortalsforBlue-CollarWorkers
Sudam Fegade, Priya Biradar, Vivekanand Dukare, and NiveditaRawate(2022)presentedajobportalspecifically designed for blue-collar professionals to address employment challenges faced by semi-skilled and uneducatedworkers.Theplatformenablesjobposting,job search, application tracking, and recruiter–job seeker interaction through a simplified mobile application. The study emphasizes reducing time, cost, and accessibility barriersforworkersfromsmalltownsseekingemployment in urban areas. While the system effectively improves outreach and usability, it relies on conventional matching techniques and lacks intelligent automation, personalized job recommendations, and AI-driven decision-making features.
3.2IntelligentJobRecommendationSystems
Priyanka Singla and Vishal Verma (2025) introduced an intelligentjobrecommendationsystembasedonsemantic embeddings and machine learning techniques. Their proposedhybridapproachperformsbi-directionalmatching betweenjobseekers’CVsandjobdescriptions,overcoming
International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056
Volume:12Issue:12|Dec2025 www.irjet.net p-ISSN:2395-0072
issues such as static and false recommendations found in traditional systems. The methodology employs natural language processing techniques, Sentence-BERT (SBERT), cosine similarity, and clustering algorithms to generate accurateandexplainablejobrecommendations.Thesystem was validated using multiple real-world datasets, demonstrating improved robustness and adaptability. Despiteitseffectiveness,theauthorsnotechallengesrelated to computational complexity and limited accessibility for non-technicalorlow-literacyusers.
3.3Voice-EnabledHuman-ComputerInteraction
PrachiTiwariandMeetkumarPatel(2024)proposedanAIbaseddesktopvoiceassistantdevelopedusingPythonand open-source artificial intelligence libraries. Their work focusesonenhancinghuman–computerinteractionthrough voice-driven task automation, information retrieval, and personalized assistance. The system integrates speech recognition, natural language understanding, dialogue management, and task execution modules using libraries suchasSpeechRecognition,NLTK,andPyTorch.Theauthors evaluatedthesystembasedonaccuracy,responsetime,and user satisfaction, demonstrating reliable performance in desktop environments. However, the study highlights limitationsinscalability,contextualintelligence,anddeeper personalization, indicating scope for integrating more advancedlearningmechanisms.
3.4 Conversational AI and Chat bots for Enhanced WebsiteUserExperience
ManojKumarDobbala and ManiShankar SrinivasLingolu (2024) studied the adoption of conversational AI technologies such as NLP, NLU, and machine learning to improveuserexperienceandengagementonwebsites.Their research highlights major benefits including personalized assistance,reducedwaittime,andcontinuousaccessibility across sectors like healthcare, finance, travel, and ecommerce. However, the authors identified challenges includingambiguityinuserintent,theneedforcontextual understanding, privacy concerns, and system integration complexities. Solutions such as advanced NLP algorithms, secure data handling, and user feedback loops were recommended,showingapromisingdirectiontowardmore human-likeinteractionmodels.
3.5 Text-to-Speech Synthesis for Multilingual Applications
G.D.RamtekeandR.J.Ramteke(2016)proposedatext-tospeech(TTS)engineenablingspeechgenerationinEnglish, Hindi, and Marathi. Their concatenative speech synthesis approachfocusedonprosodyfeaturespitch,duration,and intensitytoenhancenaturalnessofgeneratedspeech.The systemwasevaluatedusingMeanOpinionScore(MOS)tests, whichshowedlisteningqualitybetweenfairandgoodand naturalnessrangingfrompleasanttoslightlypleasant.While
effective for multilingual support and accessibility (e.g., visually-impaired users and education), the approach depends on manual corpus creation and is vulnerable to noise during voice recording, limiting scalability and adaptationtolargedatasets.
3.6 AI-Enabled Text-to-Speech for Indian Unicode Languages
ChandamitaNathandBhairabSarma(2024)developedan AI-enabled TTS system for Assamese, addressing the complexityofIndiclanguagestructures,grapheme-phoneme mapping, and Unicode text processing. They compared multipleapproachesincludingDictionary-based,HMM,CNN, andGrapheme-to-Phoneme(G2P)andfoundG2Ptobemost accurate, achieving around 89% accuracy with improved phonemealignment.Thesystemwastrainedusingacustom dataset exceeding 40,000 speech samples and utilized Tacotron2 with the Griffin-Lim algorithm for waveform reconstruction. Despite promising results, the system’s performance is constrained by dataset size and pronunciationvariationsacrossdialects,indicatingfurther needforrobustlanguagemodeling.
3.7ConversationalAIAdvancementsandChallenges
VIKASHASSIJA(2023)providedacomprehensivesurveyon the evolution and capabilities of large language models, particularly ChatGPT. Their work discusses transformer architecture, training processes, and multiple application domains including customer service, translation, summarization, and human-AI collaboration. The authors highlighted major issues such as bias, response quality control, and ethical deployment constraints, emphasizing thatdespitetherapidgrowthofconversationalAI,practical scalability and domain-specific adaptation remain open challenges.
3.8AI-DrivenCareerRecommendationSystems
Prof.KarishmaMisal(2025)proposedanAI-drivensystem forcareerpathguidancetailoredtostudentsaftersecondary education.Byusingstudentscoresandextracurriculardata, the model applies classification techniques like KNN and SVMto recommendsuitable careerdomains. Thesolution aims to minimize misaligned academic decisions and enhance personalized counselling. However, the system depends heavily on quality of training data and may not captureemotionalormotivationalfactorsinfluencinglongtermcareerchoices.
3.9Blue-CollarWorkforceandCompensationAnalysis
Ms. Shriya S and Mr. K. Gunashekar (2020) conducted an exploratory study on compensation trends among bluecollar employees in small and medium enterprises in Bengaluru. Using primary survey data, they identified concernssuchaslowwages,lackofsocialsecuritybenefits
International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056
Volume:12Issue:12|Dec2025 www.irjet.net p-ISSN:2395-0072
(ESI & PF), job insecurity, and high occupational risks despitestronglaborcontributions.Theresearchreinforced the need for structured reward policies and welfare programstoimprovemotivation,safety,andcareerstability amongblue-collarworkers.
3.10CollaborativeFilteringforJobRecommendation
Zhangetal.(2014)introducedanitem-basedcollaborative filtering technique using weighted co-apply behavior to recommend relevant jobs to candidates. Their model significantly improved precision and recall scores in job matching efficiency. The approach focuses on behavioral similarityratherthantextualattributesofjobdescriptions. However,theirsystemisaffectedbycold-startissuessince recommendationsrelyonuserinteractionhistory.Thestudy showspotentialforenhancingblue-collarjobdiscoverywith adaptivedata-driventechniques.
3.11AutomatedJobAllocationforInformalWorkforce
Amirnenietal.(2016)developedajoballocationsystemfor blue-collar workers using Aadhaar-linked identity verification and rule-based matchmaking. The system automateslabordistributionbasedondemand,availability, and skill requirements. It improves fairness and reduces dependency on middlemen in informal hiring processes. Nevertheless,theplatformstillrequiresdigitalliteracyand lacksvoice-basedsupportfornon-educatedworkers.Future enhancements should incorporate real-time decision supportformigrantlaborers.
3.12BilingualAssameseText-to-SpeechSystem
Sharma et al. (2015) explored speech synthesis for Assamese–Englishmixedlanguageusingunitselectionand HMM methods. The research improves speech continuity andintonationwhenshiftingbetweenregionalandEnglish words.Itsapplicationiscrucialforaccessibilitytechnologies inmultilingualsocietieslikeIndia.However,largeannotated datasets are required to expand vocabulary and dialect coverage. It highlights the importance of localized voice systemsforpublicservicesincludingjobsearch.
3.13IndianEnglishSpeechAdaptationforTTS
Mahantaetal.(2016)modifiedCMUphonememappingto address Indian English pronunciation variations. Their adaptation reduced mispronunciations, making speech output clearer and more natural for Indian listeners. This researchisvitalforvoice-basedsystemstargetedatIndian job seekers, reducing comprehension barriers. However, prosody and conversation-level expressiveness remain limited. Further improvements are needed for smooth interactioninreal-worldjobplatforms.
3.14DecentWorkConditionsinApparelIndustry
3.15 Physically Demanding Blue-Collar Jobs Motivation&Challenges
Chen et al. (2017) compared satisfaction levels between blue-collarandwhite-collaremployeesinChinesegarment manufacturing. Blue-collar workers reported higher dissatisfactionregardingwages,careergrowth,andsafety standards.Thishighlightstheurgentneedfortransparent jobsystemsthatprotectlaborrights.Theirstudysuggests integratinglong-termcareersupportforworkforcestability. Digitalplatformsmustincludesocialcomplianceanddecent workindicatorsforbetterjobmatching. Venancioetal.(2024)examinedphysicallydemandingbluecollar roles using 15 participants in the Philippines. The findings show workers continue such jobs mainly due to financialresponsibility,familyneeds,andsocialbelonging. Despite chronic stress, physical strain, irregular sleep patterns,andlowsalaries,workersremainmotivatedbyjob security and workmate support. The study indicates that mental well-being and work–life balance are severely affected. This underscores the necessity for digital job platforms that prioritize welfare, accessibility, and safer employmentopportunities.1
4. Comparative Discussion
Based on the review, existing research contributes significantlytowardenhancingemploymentaccessthrough different technological and social domains. AI-based job recommendation studies in Research offer intelligent matchingapproachesbutremaindependentontextualinput anduserinteractionhistory,whichlimitstheiraccessibility forlowliterateworkforcegroups.Likewise,Aadhaar-enabled joballocationmodelssuccessfullyautomatehiringforblue collarworkersbutdonotintegratespeech-basedinteraction, making them less user-friendly for digitally inexperienced populations.
ConversationalAIandmultilingualTTSsystemscontribute towardaccessibilitybyenablingassistancethroughnatural speech, yet they are rarely integrated into job-specific workflows.Researchaddressingworkerwelfarehighlights major challenges such as job insecurity, unfair wages, and high physical strain, but lacks practical implementation throughmoderntechnologies.Overall,thereviewedstudies show progress in isolated areas recommendation, voice interaction, or labour welfare but none of the approaches provideaunifiedsolutionthatensuresjobdiscovery,literacy support,andsocialwell-beingsimultaneously.
Therefore,thereexistsastrongneedforaplatformcapableof combining AI-driven recommendations, regional-language support, and blue-collar workforce empowerment into a single,inclusiveemploymentecosystem.
International Research Journal of Engineering and Technology (IRJET) e-ISSN:2395-0056
Volume:12Issue:12|Dec2025 www.irjet.net p-ISSN:2395-0072
4. CONCLUSIONS
ThisreviewconcludesthatcurrentjobportalsandAI-based systems remain largely skewed toward white-collar and educated job seekers. Despite advancements in conversational AI, voice assistance, and heuristic-driven workforce allocation, digitally marginalized workers continue to struggle with barriers such as non-English interfaces, complex app usage, and limited job security. Incorporating Indian language speech interfaces into recruitmentsystemscansignificantlyimproveusabilityfor informalsectorworkersandindividualswithminimaldigital exposure.
Additionally,consideringdecent-workindicatorssuchasfair compensation, job stability, and worker well-being is essentialtoensurelong-termadoptionandmeaningfulsocial impact.The findings stronglyjustify the development of a voice-enabled AI job portal tailored specifically for blue collarandsemi-skilledworkersespeciallyingrowingmetro regionslike Nagpur ensuringequal accessto employment opportunitiesforall.
5. REFERENCES
[1] S. Fegade, P. Biradar, V. Dukare, and N. Rawate, “Job Portal for Blue Collar Professionals,” TechRxiv Preprint, Mar. 2022, doi: 10.36227/techrxiv.19425188.v1.
[2] P. Singla and V. Verma, “An Intelligent Job Recommendation System based on Semantic Embeddings and Machine Learning,” Journal of InformationSystemsEngineeringandManagement,vol. 10,no.5s,2025.
[3] P. Tiwari and M. Patel, “Online AI Based Voice Assistant,”Article,Mar.2024.
[4] M. K. Dobbala and M. S. S. Lingolu, “Conversational AI andChatbots:EnhancingUserExperienceonWebsites,” AmericanJ.Comput.Sci.Technol.,vol.7,no.3,pp.62–70,2024.https://doi.org/10.11648/j.ajcst.20240703.11
[5] G. D. Ramteke and R. J. Ramteke, “Text-To-Speech Synthesizer for English, Hindi and Marathi Spoken Signals,”Br.J.Appl.Sci.Technol.,vol.15,no.3,pp.1–16, 2016.DOI:10.9734/BJAST/2016/24869
[6] C. Nath and B. Sarma, “AI Enabled Text-to-Speech SynthesisforUnicodeLanguage,”IndianJ.Sci.Technol., vol. 17, no. 42, pp. 4454–4461, 2024 https://doi.org/10.17485/IJST/v17i42.2645
[7] Vikas Hassija, "Capabilities, limitations, and future directions a comprehensive survey,” IEEE Access, 2023.
[8] K.Misal,S.Vaidya,A.Adpawar,S.Pandey,andP.Sarve, “AI-Driven Career Path Finder & Navigator Revolutionizing Career Guidance with Artificial Intelligence,”Int.J.Innov.Res.Technol.,vol.11,no.12, pp.6224–6233,May2025.
[9] S. Shriya and K. Gunashekar, “A STUDY ON COMPENSATION AND REMUNERATION OF BLUE COLLARED EMPLOYEES IN SMALL & MEDIUM ENTERPRISES IN BANGALORE: AN EXPLORATORY STUDY,”Int.J.Innov.Res.Manage.Stud.,vol.4,no.12, pp.191–196,Aug.2020.
[10] Y. Zhang, C. Yang, and Z. Niu, “A Research of Job Recommendation System Based on Collaborative Filtering,”in2014SeventhInternationalSymposiumon Computational Intelligence and Design, Hangzhou, China,2014,pp.533–538.
[11] S. Amirneni, M. P., J. Balaji, and N. Nivedha, “Demand Based Blue Collared Job Allocation Using Multiple Priorities and Heuristics,” in 2016 IEEE International ConferenceonKnowledgeEngineeringandApplications, Bali,Indonesia,2016,pp.226–230B.
[12] Sharma, P. Sarma, and D. Dutta, “Development of Assamese–Englishbilingualtext-to-speechsystemusing HMM-basedsynthesis,”2015.
[13] S. Mahanta and R. Sharma, “Indian English phoneme adaptationforimprovedspeechsynthesis,”2016.
[14] C.Chen,P.Perry,Y.Yang,andY.Cheng,“Decentworkin theChineseapparelindustry:Analysisofblue-collarand white-collarworkers,”Sustainability,vol.9,no.8,pp.1–15,2017.DOI:https://doi.org/10.3390/su9081344
[15] F.A.Venancio,J.A.D.Quinte,andB.T.S.Sengco,“BlueCollarWorkers:StudyonPhysicallyDemandingJobs,” PsychologyandEducation:AMultidisciplinaryJournal, vol. 18, no. 3, pp. 264–276, 2024, Doi: 10.5281/zenodo.10875909.