Hi, I'm Aditya.
CS @ Carnegie Mellon.
I build, I compete, I ship.
Carnegie Mellon University
B.S. Computer Science · Class of 2029
ICPC World Finals 2026 Qualifier · Discrete Math, Logic & Systems
Professional Experience
Where I've worked
Vidi
Current- Architected and shipped a production REST API syndicating Vidi's events to ILoveQatar (ILQ), one of Qatar's largest lifestyle platforms — 2 authenticated endpoints serving a standardized 15-field JSON schema, secured with SHA-256-hashed, scope-based API keys supporting instant revocation; verified across the full auth matrix (401/200/404).
- Designed and deployed a zero-downtime MySQL schema migration (7 additive columns, 2 new tables, automated slug backfill with collision handling) extending a live production database for third-party syndication with zero breaking changes to the existing mobile app and partner flows.
- Built a production-hardened PHP signup backend with defense-in-depth abuse prevention — IP-based rate limiting, honeypot bot filtering, silent duplicate handling, and Cloudflare-aware geo/IP capture — plus fire-and-forget webhook notifications and dual JSON/form-encoded request handling, integrated live with a Framer landing page.
- Extended the PHP affiliate portal end-to-end, adding 10+ new event-authoring fields and a dynamic JavaScript pricing-tier builder that flows partner input straight through to the public API response — closing the data-capture gap that gated the ILQ go-live.
Carnegie Mellon University
Current- Architected and shipped a full-stack course-planning web application from prototype to production over 116 commits, delivering an interactive catalog of 1,700+ Carnegie Mellon courses mapped to degree requirements across multiple majors and 15+ minors.
- Built a modular Flask + SQLAlchemy REST API (11 endpoints) with Google OAuth2 single sign-on and role-based access control spanning 4 user roles (student, advisor, associate area head, admin), enforcing permissions through reusable decorators with request-scoped caching.
- Automated a daily data pipeline that scrapes CMU's Schedule of Classes, normalizes 1,700+ course records, and tags them by campus (Pittsburgh/Qatar) while keeping the live catalog continuously in sync with zero manual updates.
- Engineered two production workflows end-to-end — a faculty course-flagging system with an admin review queue and a student course-wishlist — backed by an auditable data model that snapshots submitter identity for traceability.
- Wrote 87 automated tests (40 backend, 47 frontend) and led a post-deployment security audit that surfaced and fixed 4 authentication bugs, hardening OAuth and role-promotion logic before launch.
- Deployed a decoupled production architecture — static SPA frontend on GitHub Pages, Flask API and PostgreSQL on Render — with environment-based secret management and CORS configuration.
Mercor
Current- Synthesizing high-complexity Olympiad-level mathematical reasoning datasets to benchmark and fine-tune Frontier LLMs (GPT-4, Gemini-class).
- Engineering rigorous, human-verifiable chain-of-thought (CoT) solutions for IMO-style proofs used as gold-standard evaluation benchmarks.
- Focusing on long-horizon reasoning and proof construction to push the boundaries of model abstraction.
Olympiad Training & Competition
- Dedicated an intensive year to competitive programming and algorithmic training.
- Gold Medalist at QCPC — 1st place among 133 teams nationwide.
- Silver Medalist at ACPC — qualified for ICPC World Finals 2026, representing Carnegie Mellon University.
- Bronze Medalist at INSO (International Nuclear Science Olympiad) in Malaysia.
Selected Projects
Things I've built recently
Snoonu Dispatch Engine
Hackathon — Delivery Optimization
Python, NumPy, Haversine, Event-Driven SimulationA real-time food delivery dispatch engine with multi-order batching and smart courier assignment for Snoonu (Qatar's largest delivery platform).
- Architected a heuristic scoring function optimizing travel cost, wait time, and delay penalties across 50 couriers and 50+ concurrent orders.
- Achieved a simulated batch rate of 68%, 2.3 orders/trip, and 88% on-time delivery rate — beating baseline single-dispatch by 40%.
- Implemented hot injection re-batching and a courier state machine with 5 transition states for dynamic route optimization.
Orbital Mechanics Engine
15-112 Term Project
Python, CMU Graphics, NumPy, PhysicsA custom-built physics simulation engine for the 15-112 Term Project at Carnegie Mellon, engineered to model n-body gravitational interactions with high precision.
- Architected a ground-up physics loop using semi-implicit Euler integration to maintain stable orbital trajectories and energy conservation over long simulation times.
- Implemented dynamic gravitational field rendering using
cmu_graphics, visualizing real-time potential wells and event horizons for black holes and wormholes. - Developed a procedural universe generation system that seeds unique planetary configurations and crater distributions for infinite replayability.
- Engineered valid Hohmann Transfer mechanics and thrust vectoring, validating the simulation against theoretical delta-v requirements for orbital maneuvers.
Active Debris Removal Pipeline
Machine Learning Research
PyTorch, YOLOv8, Reinforcement LearningA 3-stage ML pipeline for autonomous space debris capture using computer vision and reinforcement learning.
- Built a detection module using YOLOv8 for real-time debris identification and an RL agent for optimal pathfinding.
- Optimized trajectory reward shaping, resulting in a 25% reduction in fuel consumption in simulations.
- Benchmarked against classical computer vision baselines, demonstrating 30%+ improvement in targeting precision.
Academic Journey
Research & Learning
University of Bristol
- Authored 'Clearing the Cosmos', supervised by 3 PhD mentors; manuscript peer-reviewed.
- Architected CNN-based detection and Reinforcement Learning (RL) pipelines for autonomous debris tracking, increasing recognition precision by 30%+.
- Leveraged PyTorch and NumPy to benchmark deep learning models against classical computer vision techniques.
Stanford University (ULO)
- Selected for the prestigious Stanford ULO program — 1 of 15 selected worldwide.
- Mastered intensive Multivariable Calculus & Linear Algebra (IM051), achieving a final grade of 91/100 in the top 10% of the cohort.
NASA / IASC
- Optimized data pipelines to process 2B+ celestial detections from WISE and PAN-STARRS telescopes, flagging irregular bodies for NASA review.
- Reduced manual triage latency by 25% through standardized visual QA templates.
- Contributed to global planetary defense efforts as an active IASC Citizen Researcher.
Ministry of Education, Qatar
- Engineered a training pipeline for the top 30 nationally ranked physics students, authoring 50+ original problems modeled after IPhO standards.
- Developed computational physics simulations using Python and NumPy to visualize complex electromagnetism and fluid dynamics.
Technical Expertise
Tools of the trade
PROGRAMMING LANGUAGES
AI & Machine Learning
Development Tools
International Achievements
Winning is fun.
Silver Medal — ACPC 2025
Africa & Arab Collegiate Programming Contest
Awarded Silver in the Continental round. Qualified for the ICPC World Finals 2026, representing Carnegie Mellon University. Ranked among top teams in the region.
ViewGold Medal — QCPC 2025
Qatar Collegiate Programming Contest
Secured 1st Place among 133 teams nationwide. Demonstrated superior algorithmic problem-solving speed and accuracy to qualify for regionals.
ViewDiamond Award — ISPhO 2024
International Scientific Physics Olympiad
Best performer from the Qatar national team. Awarded 'Most Creative Answers' distinct honor for novel physics problem approaches.
ViewBronze Medal — INSO 2025
International Nuclear Science Olympiad
Won Bronze for Qatar amidst stiff international competition. Raised the national flag on the global stage for nuclear science proficiency.
ViewBronze Medal — INSO 2024
International Nuclear Science Olympiad
Secured Bronze medal in the Philippines, consistent excellence in nuclear physics theory and practical applications.
ViewNational Rank 1 — IYMC 2023
International Youth Math Challenge
Achieved the highest national rank in the final round of the IYMC, showcasing exceptional mathematical aptitude.
Reach out to me!
I'm always open to discussing new opportunities, collaborations, or just connecting with fellow engineers.
© 2026 Aditya Vivek. Built with Next.js & Three.js.