AI Engineer with less than a year in NLP & Deep Learning
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
6th-semester Information Technology student at Universitas Sumatera Utara (GPA 3.76/4.00) with hands-on experience in Backend Development and AI/Machine Learning. Currently serving as an AI Engineer in a nationally competitive coding program, with practical exposure to NLP, deep learning, and end-to-end model development from architecture to deployment. Proficient in Python, PHP/Laravel, and JavaScript, and experienced in collaborating across multidisciplinary teams to deliver data-driven, real-world solutions. Seeking an internship to further apply these skills in a product-driven environment.
Universitas Sumatera Utara
Bachelor of Information Technology
August 1, 2023 – Present
DBS Foundation
AI Engineer Cohort
February 1, 2026 – June 30, 2026
India
CareerLens - AI Based Career Recommendation System
June 1, 2026 – June 30, 2026
Designed and implemented a FeedForward Neural Network with Cosine Similarity to match user profiles against 68 job roles across three input pipelines: career interest selection, skill tag matching, and RIASEC aptitude assessment. Achieved ~86% accuracy on the SkillModel (trained on 30.000 generated samples) and ~96% accuracy on the RIASECModel (trained on 5.000 samples), evaluated using Accuracy, F1-Score, Precision, Recall, and Precision@K. Owned the full model lifecycle architecture design, data preprocessing, custom training loop with tf.GradientTape, inference logic, and deployed the AI service via Hugging Face Spaces integrated with a FastAPI backend. Resolved 5+ critical model bugs including binary label correction, custom callback incompatibility with TF 2.20, and cosine similarity output normalization, improving recommendation reliability across 68 career classes. Applied Precision@K evaluation, dropout regularization, and dimensionality capping (390-dim -> 128-dim hidden layers) to balance model performance and generalization across a dual-output architecture (role score + skill gap).
View ProjectShopee Review Sentiment Analysis
May 1, 2026 – May 31, 2026
Built an end-to-end Indonesian-language sentiment analysis pipeline on Shopee Play Store reviews, comparing 3 model architectures: LSTM + TF-IDF (80,90%), BiLSTM + Word2Vec (91,59%), and CNN-BiLSTM + Keras Embedding (97,55% test accuracy). Achieved 97,55% test accuracy with the best-performing CNN-BiLSTM architecture using multi-scale parallel convolutions (kernel sizes 3, 4, 5), bidirectional LSTM, and dual pooling (Average + Max) for richer text representation. Implemented a lexicon-based labeling strategy using an Indonesian sentiment dictionary (InSet Lexicon) with negation handling, replacing star-rating-based labels to improve annotation quality on 3-class classification (positif, netral, negatif). Applied 8-step NLP preprocessing pipeline for Bahasa Indonesia including Sastrawi stemming, slang normalization (50+ slang terms), stopword removal, and tokenization alongside RandomOverSampler to handle class imbalance across all 3 models.
View ProjectLibrary Management System – SMAN 2 Binjai
December 1, 2024 – December 31, 2024
Delivered both Front-End and Back-End features including book catalog, member management, borrowing & return tracking, fine management, and loan extension requests. Implemented 3-tier role-based access control (Admin, Staff, Member) with secure login and profile management for each user level. Administered MySQL database schema design and query optimization to support structured and efficient data storage.
View ProjectMuzzy - Semantic Music Discovery Website
December 1, 2024 – December 31, 2024
Created a semantic web-based music platform as a 7-person team for a Web Semantics course final project. Integrated SPARQL query integration with Apache Jena Fuseki as the RDF triplestore, enabling semantic search across song titles, artists, albums, genres, labels, writers, and producers. Developed front-end features including song detail pages, lyrics display, and in-browser audio playback using PHP and EasyRdf library.
View ProjectAdvanced Deep Learning Projects
Dicoding Indonesia
January 1, 2026 – Present
Fundamentals and Applications of Generative AI
Codepolitan
January 1, 2026 – Present
Data Analysis with Python
Cognitive Class (IBM)
January 1, 2025 – Present
The candidate scored 88% on the Data Scientist — Artificial Intelligence test, indicating a strong grasp of core AI/ML concepts, deep learning, and practical application, aligning well with the target role.
Cultural Fit Analysis
The candidate's participation in a competitive AI Engineering training program and diverse project portfolio (sentiment analysis, recommendation systems, semantic web, library management) demonstrate initiative, a strong learning drive, and adaptability. The academic project 'Muzzy' shows an ability to collaborate in a team setting. The low psychometric score, however, raises a flag regarding potential challenges in areas like team collaboration or stress handling, which are important for cultural integration and effective teamwork. The candidate's focus on AI/ML projects aligns well with an AI Engineer role.
Soft Skills & Operational Fit
The candidate's project descriptions indicate an ability to work in cross-functional teams (CareerLens) and manage project deliverables. The organizational experience as Vice Secretary suggests good coordination and documentation skills. However, the psychometric test score (196/500) is low, which might indicate potential areas for development in logical reasoning, work attitude, stress handling, or team collaboration, which are critical for operational fit in a senior role. Further investigation into the psychometric results would be beneficial.
Strengths
Limitations