Hello, I'm
Aditya Joshi
Computer Science Student at NUS | Artificial Intelligence
Hello, I'm
Computer Science Student at NUS | Artificial Intelligence
Get To Know More
Member of Technical Staff Intern at The OpenAI Deployment Company Present
AIoT Software Developer Intern at Bosch Digital
Creative Technologist at Serial Communication Private Limited
R&D Software Developer Intern at Laconic Technology
Tech Executive at Artimal
Bachelor of Computing in Computer Science
Artificial Intelligence
National University of Singapore
Aug 2023 - May 2027 (expected)
Hi, I'm Aditya! I'm a Computer Science student at NUS specializing in Artificial Intelligence and a Member of Technical Staff Intern at The OpenAI Deployment Company in Singapore. I build multi-agent AI workflows, enterprise applications, computer vision systems, and backend software. My experience spans Singapore and an internship in Bangkok through NUS Overseas Colleges Thailand, where I built 100+ REST APIs and a RAG data assistant. Beyond software, I led publicity at NUS Electronic Music Lab and volunteered with Tara Baal Sanstha to support children's education. Outside of work and studies, I enjoy music production, filmmaking, poetry, basketball, hip-hop, anime, and photography. I speak English fluently; Hindi is my native language.
Explore My
The OpenAI Deployment Company, Singapore | Jun 2026 - Present
Co-designed and built a Python-based, spec-driven AI workflow that converts business requirements into user stories, implementation plans, and working applications in 16-24 hours, reducing development cycles that previously took weeks. Implemented multi-agent planning, coding, review, and validation workflows using Codex CLI, YAML/JSON workflow state, shell tooling, and Git/GitHub; adopted by 10+ internal users. Built human-in-the-loop review checkpoints and automated evaluation loops to surface blockers, incorporate reviewer feedback, and improve agent prompts and orchestration.
Bosch Digital, Robert Bosch (SEA) Pte Ltd, Singapore | Jan 2026 - Jun 2026
Owned an Angular micro-frontend that embedded external web pages within an enterprise AIoT platform and was used in client demonstrations. Built 30+ Playwright end-to-end tests for core search and file-upload workflows, catching major defects before release, and created and maintained three Azure DevOps pipelines triggered after merged pull requests. Developed a Keycloak-secured Python API for an RPA bot and deployed scheduled report-download automation on a Windows VM, removing recurring manual intervention.
Serial Communication Private Limited, Singapore | Oct 2025 - Jan 2026
Benchmarked candidate YOLO architectures, selected YOLOv11m-OBB, and trained two model iterations on 10K+ images across 20+ product categories using an 80:20 dataset split, reaching 12 FPS in prototype inference. Built an Ultralytics/CVAT and Label Studio pipeline with prediction APIs and OBB-to-JSON pre-annotations, reducing annotation time from 6 to 2 hours per 1,000 images.
Laconic Technology, Bangkok, Thailand | May 2025 - Aug 2025
NUS Overseas Colleges Thailand
Owned backend development in a four-person intern team, implementing 100+ Java/Spring Boot REST APIs and PostgreSQL schemas for POS and e-commerce workflows; delivered the backend to the internal engineering team in under three months. Built a Python data assistant with RAG, Vanna, and Llama 3 for natural-language queries over structured data. A management demonstration led to the approach being selected for further use.
Artimal, Singapore | Oct 2024 - Jan 2025
Built mascot chatbots and MySQL data workflows, delivering the project with a cross-functional team in under three weeks.
Bookism, Khatima, India | Jan 2022 - Sep 2022
Co-founded book rental start-up, gathered 200 initial users within a week, rented 30+ books, and conducted 3 seminars onboarding 100+ students.
NUS Electronic Music Lab | Aug 2024 - May 2025
Led digital publicity campaigns that increased online engagement by 57% and event attendance by 70%.
Tara Baal Sanstha | Sep 2021
Taught 20 underprivileged children, focusing on literacy, reading, and academic support. Donated books and learning materials to support long-term educational development.
Browse My Recent
Real-time gesture-controlled musical instrument | Sep 2025
Built a real-time musical instrument using MediaPipe and OpenCV, tracking 21 hand landmarks at 30+ FPS and mapping seven gestures to chords across 12 musical scales. Developed threaded audio synthesis with Python and SoundDevice, generating harmonic chords from layered sine waves with instant scale and chord-duration adjustments and low-latency playback.
Full-stack ML app for YouTube comment sentiment | Sep 2025
Trained an XGBoost sentiment classifier on 50K IMDB reviews using spaCy and 5K TF-IDF features, achieving 85% accuracy. Built a React/Flask app with YouTube Data API integration to analyze up to 1,000 comments per video within seconds and surface sentiment distributions and high-confidence examples.
Machine Learning system predicting US flight delays with 77% accuracy
Built a comprehensive ML pipeline using XGBoost on over 558K flight records to predict delays. Features a React frontend with responsive UI, a Flask backend serving predictions, and a multi-model architecture for distance, duration, and delay probability estimation. Provides real-time delay predictions with feature importance insights.
Java-based lossless file compressor using Huffman coding with CLI and Maven packaging
Implemented a complete Huffman compression pipeline: frequency analysis, priority-queue tree construction, code generation, and bit-level I/O with a custom file format. Built a professional CLI supporting compress/decompress operations and packaged it as an executable JAR via Maven for cross-platform usage.
Custom Tic Tac Toe variant with AI and move-removal mechanics
Designed and developed a unique Tic Tac Toe variant with move-removal mechanics, an AI opponent using Negamax with alpha–beta pruning, and a responsive Streamlit web interface with real-time updates.
Machine Learning model predicting global video game sales
Developed a machine learning model to predict global video game sales using Linear Regression. Visualized model predictions against actual sales with scatter plots to improve understanding of model performance.
RAG document platform with Apollo 11 recognition | May - Aug 2024
Built the Flask backend for a two-person RAG platform using Gemini, LangChain, Sentence Transformers, FAISS, and MongoDB to extract propositions and retrieve the top five semantic matches from uploaded documents. Exposed file-upload and query APIs for a React frontend. Earned Apollo 11, the second-highest recognition in the NUS Orbital Programme.
Hybrid CLI-GUI candidate and job role management system
Built a candidate and job role management system tailored for HR professionals and recruiters. Designed and implemented a hybrid CLI–GUI interface using Java and JavaFX, enabling efficient text-based commands with visual feedback.
Get in Touch
