PLACIFY AI : AI-POWERED PLACEMENT PREPARATION PLATFORM WITH SANDBOXED CODE EXECUTION, RETRIEVAL-AUGMENTED MENTORING AND PLACEMENT-READINESS PREDICTION

Placify AI shows that a placement-preparation platform can integrate coding practice, learning, interview rehearsal and resume feedback while adding machine-learning guidance that is transparent about its limits. Separating a FastAPI intelligence service from the Express transactional API, isolating untrusted code in per-submission containers, and giving every AI feature a deterministic fallback make the system safe to run and easy to evolve. On held-out data the readiness model reached 0.860 ROC-AUC (at the ceiling of its noisy labels) and the difficulty classifier 76.7% accuracy; retrieval placed the expected chunk first for 83.3% of queries; the ML service passed all 103 of its tests and the sandbox all 60 of its checks. The evaluation also exposed concrete weaknesses: keyword stuffing, limited reference coverage for interview scoring, an oversized problem-list payload, and inconsistent test data in part of the problem bank. The readiness score is trained on synthetic data and must be re-fitted and validated on real placement outcomes before it is used to advise students or institutions.

[1] React Documentation, "React," Meta Open Source. Available: https://react.dev/ [2] Express Documentation, "Express - Node.js web application framework," OpenJS Foundation. Available: https://expressjs.com/ [3] Prisma Documentation, "Prisma ORM," Prisma Data, Inc. Available: https://www.prisma.io/docs [4] FastAPI Documentation, "FastAPI - Modern Web Framework for Building APIs with Python." Available: https://fastapi.tiangolo.com/ [5] F. Pedregosa et al., "Scikit-learn: Machine learning in Python," J. Mach. Learn. Res., vol. 12, pp. 2825-2830, 2011. [6] J. H. Friedman, "Greedy function approximation: A gradient boosting machine," Ann. Statist., vol. 29, no. 5, pp. 1189-1232, 2001. [7] G. Salton and C. Buckley, "Term-weighting approaches in automatic text retrieval," Inf. Process. Manage., vol. 24, no. 5, pp. 513-523, 1988. [8] A. McCallum and K. Nigam, "A comparison of event models for naive Bayes text classification," in AAAI-98 Workshop on Learning for Text Categorization, 1998. [9] N. Reimers and I. Gurevych, "Sentence-BERT: Sentence embeddings using Siamese BERT-networks," in Proc. EMNLP-IJCNLP, 2019, pp. 3982-3992. [10] W. Wang et al., "MiniLM: Deep self-attention distillation for task-agnostic compression of pre-trained transformers," in Proc. NeurIPS, vol. 33, 2020.