Project Info
Inspiration
For many college students, aptitude is an integral part of the placement preparation. However, most students do not have an effective way of practicing and preparing while understanding what companies care about, what they are weak in, and what they should focus on next. We wanted to build something that helps students get more out of their placement preparation by making it more personalized, data driven and consistent This inspired us to build AptiForge, an AI powered placement aptitude preparation platform that bridges the three important pillars of preparation: company tendencies, personal performance analytics, and daily practice. Our aim was simple: do not just practice more, practice better
What it does
AptiForge is an AI powered platform that helps students prepare better for their placement aptitude exams. AptiForge consolidates multiple experiences of preparation into one single place, namely: Comprehensive Aptitude: Covers QA, LR, VA, and Technical aptitude across various levels and types of questions Company Intelligence: Use historical and public information on question patterns of companies to understand what topics to expect from a certain company Data Driven: See analytics on what topics are important, what difficulty levels are common, and what are the patterns in specific companies Personalized Performance Analytics: See your own performance, accuracy, speed, topics of strength and weakness, and patterns in mistakes AI Powered Mistake Mining: Understand what type of mistake you commonly make (conceptual, calculation, misinterpretation, etc) Personalized Recommendations: Get personalized recommendations based on your performance, weaknesses, target companies, and learning history Daily 10: Practice a personalized set of 10 aptitude questions everyday Learning Reinforcement: Learn topics through repeated practice in the upcoming Daily 10s of the next few days Streaks, Challenges, and Contests: Keep track of your streaks, and participate in daily contests to keep yourself motivated Placement Readiness: Get an estimate of how well you are prepared for a particular company based on your accuracy, speed, topic strengths, recent practice, and importance of topics in the company AI Mentor: Ask your mentor questions and get personalized help with your performance, weaknesses, and recommendations The learning loop in AptiForge follows this cycle: [learn → practice → analyze → understand weakness → personalize → practice again] AptiForge aims to take aptitude preparation from being a set of random questions to a personalized and consistent learning experience that follows you for your entire placement journey.
How we built it
AptiForge is a state of the art full-stack AI platform built to offer the best possible learning experience for placement aptitude preparation. The frontend is written in Next.js, React, TypeScript, Tailwind, and shadcn/ui . We use prisma + sqlite for our database, which stores: students questions topics companies practice data user responses learning data daily 10s streaks and contests performance analytics We use recharts to render the various charts related to company data, and students' performance analytics. Our intelligence layer is powered by powerful GPT-5.6, which is used for various purposes within the application, such as: explaining questions analyzing mistakes identifying weaknesses generating recommendations providing learning insights helping with the AI mentor feature The personalization engine uses a mix of deterministic and AI driven insights to power various recommendations. We have built our own personalization engine for recommending topics based on a mix of factors such as: weak topics recent mistakes topics learned recently target company difficulty of the question historical performance and streaks practice frequency Similarly, the Daily 10 feature uses these signals to recommend a personalized set of 10 questions. For example, if you learn Percentages today, then the Daily 10 for tomorrow can focus on Percentages along with your weak topics and recent mistakes. For company patterns, we have a collection of curated data that has been mined from public sources and historical information on company placements. We analyze this data for patterns in topics and difficulty and use that to power company specific analytics. We make sure to separate between historical data, curated data, and AI generated data, and make this clear to the user. Similarly, we use Codex extensively to accelerate our development process, and power various aspects of the full-stack application, such as: project scaffolding frontend development database modeling API dev AI integration debugging, testing, and iteration
Challenges we ran into
One of our main challenges was to make sure that the platform provides a truly personalized experience based on the user's performance, and not just randomly generated questions. Aptitude has a huge number of topics and question types, and we had to come up with an intelligent way to personalize the practice sets based on what the user practices and learns. Another challenge was to make sure that while using GPT-5.6 for recommendations, the application remains deterministic and the scoring, streaks, readiness, and other factors remain reliable and predictable. We had to make intelligent choices in separating the roles of AI and deterministic code to make the system reliable and yet personalized. A challenge specific to the company patterns was to make sure that we differentiate between historical data, curated data, and AI generated data, and not present it in a way that implies anything other than what it is. Finally, we had to make sure that we deliver a complete experience within the limited hackathon time by focusing on the most impactful features
Accomplishments we're proud of
We are extremely proud to have built AptiForge as a personalized aptitude preparation platform. Our main accomplishment is to make sure that the application provides a continuous loop of learning, practice, and personalization based on the user's performance. We are especially proud of having built: a comprehensive preparation experience covering multiple areas of aptitude company specific preparation and analytics personalized Daily 10 that builds upon what the user learns everyday learning analytics that goes beyond simple performance percentages "Mistake DNA" to understand and explain weaknesses in the user's performance placement readiness estimation based on performance, speed, and topic importance in target companies. having streaks, challenges, and contests as part of the motivational system integration of GPT-5.6 into the learning experience and using it for recommendations, explanations, and the AI mentor the use of codex to accelerate the development of the full-stack application The type of experience that we wanted to build was simple: a personalized aptitude preparation where AptiForge knows what you learned, what you struggle with, what companies care about, and helps you decide what to practice next
What we learned
Building AptiForge was an eye-opening experience for us, and we learned a lot of important lessons. One important lesson was that when it comes to education, personalization is key and AI should be used to drive personalization. We learned that the most useful AI assistant is the one that knows the most about the user. We realized that an effective AI learning assistant is a combination of: AI Intelligence personal performance data learning history and feedback from the user We also learned the importance of separating between AI generated data and other data sources, and making sure to cite information when appropriate. For example, while AI can power recommendation and explanations, the system should be deterministic when it comes to scoring, streaks, placement readiness, and other important factors. We also learned the usefulness of Codex in rapidly developing a full-stack application. Instead of spending too much time on implementing generic functionality, we were able to use Codex to accelerate the development and focus on more important parts of the application. Most importantly, we learned that personalization is continuous. A student should be able to get recommendations based on their learning, mistakes, and progress. That is why some of the most impactful features in AptiForge are the continuous streaks, learning, and practice loops.
What's next
While AptiForge focuses on aptitude, our ultimate goal is to transform it into a comprehensive AI powered placement preparation ecosystem. Some of our immediate goals include: increasing the number of questions and topics covered expanding the company specific patterns improving the company specific preparation and practice introducing adaptive mock placement tests offering personalized preparation planning based on a student's placement calendar adding coding and technical interview prep supporting multilingual practice and explanations adding voice chat for AI learning and practice enhancing the AI mentor with more long-term planning and insights adding institute wide analytics and faculty dashboards Our ultimate vision is to make sure that the students get better recommendations with every use. We hope to empower every student to answer this important question: "I have a placement test coming up. What should I practice and learn today to be better prepared for tomorrow?"
AptiForge
Forge your aptitude. Target your placement. AptiForge is a polished hackathon MVP for students preparing for placement aptitude rounds. It turns performance, company-pattern datasets, learning activity and daily habits into a focused next-practice recommendation.
What makes it different
Instead of only serving questions, AptiForge closes a loop: learn a topic → practice → track accuracy and speed → identify mistake patterns → weight company relevance → personalize Daily 10 → strengthen streak and readiness.
Included experiences: demo login/dashboard, question practice with feedback, Daily 10, readiness, streak display, Mistake DNA, learn-topic reinforcement, company intelligence/comparison, analytics, mentor fallback, and leaderboard.
Architecture
Next.js App Router provides the UI and API routes. Demo data and the deterministic personalization engine live in lib/; no external service is required for judging. A Prisma SQLite schema is included as the production migration path. AI routes safely use a deterministic response if OPENAI_API_KEY is unavailable; use OPENAI_MODEL (default gpt-5.6) when adding live OpenAI calls server-side.
Readiness is accuracy 35%, speed 20%, topic mastery 20%, recent improvement 15%, and company relevance 10%. Daily 10 prioritizes recent learning, weak topics, past mistake patterns, company relevance and revision.
Company analytics use seeded, educational sample mappings inspired by publicly available historical/candidate-reported data. They are not official hiring criteria or a hiring prediction.
Run locally
npm install
copy .env.example .env
npm run dev
Open http://localhost:3000 and choose Demo dashboard. Demo student: Yogita, readiness 78, TCS target, 12-day streak.
Validation commands: npm run lint, npm test, npm run build.
Deployment
Deploy the Next.js project to Vercel. Set OPENAI_API_KEY, OPENAI_MODEL, and a production DATABASE_URL if enabling live AI and persistence.
How Codex Accelerated Development
Codex scaffolded this Next.js MVP, implemented its seeded personalization engine and routes, built the responsive UI, added a Prisma migration schema and deterministic tests, then ran build/test checks. This log intentionally describes only work performed in this workspace.
Analysis
View
Metric
- 5
Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.
Technology
- CSSIn code
- JavaScriptIn code
- Next.jsIn code
- ReactIn code
- Tailwind CSSIn code
- TypeScriptIn code
- Node.jsClaimed
- VercelClaimed
6 of 8 appear in the indexed code. 2 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.
AI coding agents
No AI coding agent signals were found in this repository.
Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.
Codebase size
Source size
56 KB
Source files
40
Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.
Repository
Yogita-61/Apti_forge
47 files · 350 KB · @ 612c8a6
Structure
Interface
17 files · 36%Screens, components and styles rendered to the user.
+6 moreAPI & routing
11 files · 23%Request entry points: routes, handlers and controllers.
Application logic
6 files · 13%Domain rules, services and shared utilities.
Data & schema
1 file · 2%Schema definitions, migrations and data access.
Supporting
Layers are inferred from where files sit in the tree, not from reading the code. A project that names its directories unconventionally will read oddly here — open the file browser to check anything the diagram implies.
Languages
- TypeScript80%
- CSS15%
- Markdown5%
- JavaScript0%
Share of indexed source by file size. Binary and vendored files are excluded.
Dependencies
package.json
npm · 23- @prisma/client
- @vitejs/plugin-react
- autoprefixer
- bcryptjs
- class-variance-authority
- clsx
- lucide-react
- next
- postcss
- react
- react-dom
- recharts
- tailwind-merge
- tailwindcss
- zod
- +8 more
Declared in the repository’s manifests at the indexed commit. A declared package is not proof it is used, and runtime dependencies are listed first.
This project’s features have not been analysed yet.
Export this project's context (description, README, evidence, key source files) to chat with an AI agent elsewhere.