International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 13 Issue: 02 | Feb 2026 www.irjet.net p-ISSN: 2395-0072
Career Guidance & Recommendation on Placement using Machine Learning and DSA Visualizer
Jaldu Bhupesh Sai Sri1 , Malees Greeshma2 , Maddi Sateesh Reddy3
1Final Year Undergraduate Student, Dept. of Computer Science and Engineering, Joginpally B.R. Engineering College (JBREC), Hyderabad, Telangana, India
2Final Year Undergraduate Student, Dept. of Computer Science and Engineering, Joginpally B.R. Engineering College (JBREC), Hyderabad, Telangana, India
3Final Year Undergraduate Student, Dept. of Computer Science and Engineering, Joginpally B.R. Engineering College (JBREC), Hyderabad, Telangana, India
Abstract - Intherapidlyevolvinglandscapeofsoftware engineering, students face a dual challenge: mastering complex Data Structures and Algorithms(DSA) required for technical interviews and navigating an overwhelming array of career specializations. While existing solutions address these issues in isolation - either through standalone algorithm visualizers or static career counselling tools- there is a lack of integrated platforms that bridge the gap between skill acquisition and career discovery. This paper proposes a novel, dual module system. The first module is an interactive Algorithm Visualizer developed using React, capable of animating sorting, searching, and graph algorithms to enhance conceptual retention. The second module is an Ai-Job Recommendation Engine built on a Flask backend. It utilizes TF-IDF vectorization and Cosine Similarity on a dataset of over 1 million job records to match user skills withoptimaljobroles.Furthermore,thesystemintegrates Generative Ai (via Groq/Llama-3) to create personalized learning roadmaps and aggregates course resources. The results demonstrate a high-precision matching capability and an improved learning experience for students preparingfortechnicalplacements.
Key Words: Algorithm visualizer, sorting, searching, graph algorithms, Ai-Job Recommendation Engine, TFIDF vectorization, Cosine Similarity
1. INTRODUCTION
Thedemandforskilledsoftwareengineershasled to a proliferation of specialized roles, ranging from Data Science to Full Stack Development. However, a significant disconnects remains between the academic curriculum and industry requirements. Two primary hurdles exit for engineering students: the difficulty in visualizing abstract algorithmic concepts and the uncertainty in selecting a careerpaththatalignswiththeiracquiredskills.
Traditionalteachingmethodsforalgorithmsoften rely on static diagrams, which fail to convey the dynamic natureofoperationslikerecursivetreetraversalsorgraph pathfinding. Simultaneously, career guidance is often subjective, relying on human advisors who may lack real-
timedataonindustrytrends.Thisresearchpaperpresents a unified web-based platform that addresses these challengesholistically.Theproposedsystemcombines:
1.1 Visual Learning
A dynamic visualizer for arrays, linked lists, stack, queue,trees,graphs.
1.2 Intelligent Guidance
A content-based recommendation system that analyses userskillsagainstamassivedatasetofjobdescriptions to predictsuitableroles.
1.3
Actionable Roadmaps
Integration of LLMs to generate step-by-step preparationguidesfortherecommendedroles.
2. LITERATURE SURVEY
The development of this system draws upon various domains of educational technology and machine learning.
2.1 Placement Prediction Systems
Kulkarnietal.developedanalgorithmvisualizerusing React.js, focusing on sorting and pathfinding algorithms like Merge Sort, Dijkstra’s algorithm. Their research highlighted that visual information is processed faster than abstract text, significantly aiding student retention. However,theirscope was limitedto visualization without linkingtheseskillstospecificcareeroutcomes.
2.2 Placement Prediction Systems
Divya et al. proposed a placement analysis system using supervised machine learning algorithms such as SupportVectorMachines(SVM)andRandomForest.Their work focused on classifying students based on academic history(grades) to predict the probability of placement. While effective for administrative forecasting, it lacks a
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 13 Issue: 02 | Feb 2026 www.irjet.net p-ISSN: 2395-0072
mechanism to recommend specific job roles based on technicalskillsratherthanjustgrades.
2.3 Personalized Career Recommendation
Qamhieh et al .introduced a fuzzy logic-based recommender system (PCRS) for engineering students. Their approach utilized personality tests and academic performance to suggest engineering disciplines. While novel, this approach relies heavily on psychometric data rather than the explicittechnical skills andkeywords that modern applicant Tracking Systems(ATS) and recruiters prioritize
3. METHODOLOGY
The proposed system architecture is divided into two distinct but complementary modules: the Learning Module (Visualizer) and Recommendation Module(CareerAI).
3.1 Module 1: DSA Visualizer
The visualizer is built using a modern JavaScript framework to render animations of data structures. It supportsthefollowingoperations:
• Searching: Linear,Binary,Jump,Interpolation, Exponential,andFibonaccisearch.
• Sorting: Visualization of bubble, insertion, quick, and mergesortallowing users toseeelementswapping and partitioninginreal-time.
• Linear Data Structures: LinkedLists(Traversal, Insertion,Deletion),Stacks(push/pop),and Queues(Enqueue/Dequeue).
• Non-Linear Data Structures: Trees (BFS,DFS) and Graph(BFS/DFS).
Thismodulepreparestheuserforthetechnicalinterviews associatedwiththejobsrecommendedbyModule2.
3.2 Module 2 : AI Job Recommendation Engine
This module utilizes a Content-Based Filtering approach. Unlikecollaborativefiltering, which requires userhistory, this system analyses the semantic relationship between a user'sskillsetandjobdescriptions.
1.Data Collection: A dataset containing approximately 1 million job records, including titles and skill requirements,isutilized
2.Preprocessing: Job description and user inputs are cleaned(removingstopwords,andspecialcharacters).
3.Vectorization(TF-IDF): We employ the Term Frequency-inverse Document Frequency (TF-IDF) vectorizer. The static is intended to reflect how important a word (skill) is to document (job description)inacollectionorcorpus.
4.Similarity Calculation: We calculate the Cosine similarity between the user’s between skill vector (U) andthejobvectors(J)tofindtheclosestmatches.
3.3 Generative Roadmap & Course Aggregation
Once a role is predicted, the system calls the Groq API(utilizing Llama-3 models) to generate a custom markdown roadmap. Simultaneously, a web scraper fetches relevant courses from educational platforms to provideimmediatelearningresources.
4.SYSTEM IMPLEMENTATION
The system implemented using a Python-Flask backendandReact.jsfrontend.
4.1 Backend Logic (Python)
The core recommendation logic is encapsulated in train.pyandapp.py.
• Training: The train.py script reads the dataset in chunks to manage memory efficiency. It initializes a TfIDF Vectorizer with a vocabulary limit of 12,000 features to reduce noise. The resulting sparse matrix is cachedusingpickleforfastinference.
• Inference: Inapp.py,theget_recommendationsfunction transformsuserinputintoavectorandcomputescosine similarityagainstthecachedmatrix.
• Course Scrapping: courses.py uses BeautifulSoup to scrape course data. It includes a fallback. Mechanism to serve curated course lists(e.g., for “Data Scientist” or “WebDeveloper”)iflivescrapingfailsorisblocked.
4.2
Generative AI Integration
The roadmap.py module integrates the Groq client . A prompt engineering approach is used to instruct the LLM to act as a “Career Coach,” generating structured advice covering Foundation, Core Competencies, and Advanced Specialization.
4.3
Visualizer Implementation
The visualizer maintains the state of data arrays. As algorithmsexecute,statechanges(swap,comparisons)are pushed to an animation queue, which is rendered sequentially to the user, providing a step-by-step walkthroughofthealgorithm.
5. RESULTS AND DISCUSSION
5.1Recommendation Accuracy
The model was evaluated based on the sparsity of the TF-IDF matrix. With a sparsity greater than 96% , the model demonstrates a high ability to distinguish between
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 13 Issue: 02 | Feb 2026 www.irjet.net p-ISSN: 2395-0072
unique skill sets (e.g., distinguishing “React/Redux” for Frontendrolesvs.“Pandas/Scikit-learn”forDataScience).
• Input: “python,SQL,React”
• Output: Full Stack Developer, Python Developer, Data Engineer.
Thesystemsuccessfullyidentifiesrolesthatutilizethe intersectionofprovidedskills.
5.2 User Interface
The interface features a “premium dark UI” designed forvisualcomfort.Theresultssectiondynamicallyrenders the “Match Strength” (High/Medium) based on the cosine similarity score, capped at 96% to avoid overfitting illusions.
5.3 Performance
• Latency: TheuseofcachedPicklemodelallowsforsubsecondrecommendations(inferencetime<200ms).
• Scalability: The chunking method used in training allows thesystemtohandle datasets exceeding 10Lakh rowswithoutmemoryoverflowerrors.
6. CONCLUSION
This paper presented an integrated platform that merges algorithmic education with AI-driven career planning. By utilizing TF-IDF and Cosine Similarity, the system provides objective, skill-based job recommendations, overcoming the subjectivity of traditionalcounselling.TheinclusionoftheDSAvisualizer ensures that students have the immediate resources to prepare for the technical demands of their recommended careers. Future work will involve integrating a collaborative filtering mechanism to recommend jobs based on successful alumni profiles and expanding the visualizer to include complex dynamic programming problems.
REFERENCES
[1] A. Kulkarni, S. Padave, S. Shrivastava, and V. Kawtikwar, "Algorithm Visualizer," International Journal for Research in Applied Science and Engineering Technology(IJRASET),vol.11,no.VII,pp.1818-1823,July 2023.
[2]N.Divya, S.Namburu, and R.Raja,"Student Placement Analysis using Machine Learning," in Proceedings of the 8th International Conference on Communication and Electronics Systems (ICCES 2023), IEEE, 2023, pp. 10271031.
[3] M. Qamhieh, H. Sammaneh, and M. N. Demaidi, "PCRS: Personalized Career-Path Recommender System for
Engineering Students," IEEE Access, vol. 8, pp. 214039214049,2020.
[4] F. Provost and T. Fawcett, Data Science for Business, O'ReillyMedia,2013.(GeneralMLReference)
[5]J.Ramos,"Using TF-IDFtoDetermineWordRelevance in Document Queries," Proceedings of the First Instructional Conference on Machine Learning, vol. 242, pp.29-48,2003.(MethodologyReference)