ai
I built a 3D game engine with AI before Opus 4.8 and learned a lot
A while back, before Opus 4.8 dropped, I gave myself a challenge: build a real 3D game engine from scratch with AI, and see how far I could actually push it. It was never meant to ship. No product, no users, no roadmap. Just me trying to learn what building with AI really feels like when you go deep on something genuinely hard. Three weeks later I had a working 3D engine. Plenty of it broke along the way, and I rewrote things more times than I'd like to admit. But that's exactly where the learning was. AI didn't hand me an engine. It handed me a thousand decisions I still had to make and understand myself. Posting a video of it here because this is what DevConnect is for: real things people build with AI, rough edges and all. Curious from anyone who's tried something similar: what's the hardest thing you've built with AI just to learn, not to ship?
Shipped: AI-Powered Product Review Analyzer
I built an AI-powered Product Review Analyzer that turns customer reviews into useful insights. The goal was to make it easier to understand what customers actually think about a product without manually going through hundreds of reviews. What it does Analyzes product reviews Identifies customer sentiment Extracts useful insights from review data Presents the results through a simple interface What I learned 🤖 Working with AI and NLP 📊 Processing real-world review data 💻 Building an end-to-end AI application 🎨 Improving UI/UX for an AI tool 🧩 Turning an idea into a working product I'm still improving the project and would love feedback from other builders. GitHub: https://github.com/nidhishreebai-gif/product-review-analyzer What feature would you add to make this more useful? 🚀
Automated AI Lead Follow-Up System (n8n + Groq + Gmail + Sheets)
Just shipped an automated AI Lead Follow-Up System built with n8n, Groq, Gmail, and Google Sheets.
Building Jobsly: AI Job Application Autopilot
🚧 Building Jobsly — an AI Job Application Autopilot I’ve been working on Jobsly , an AI-powered job application assistant designed to make applying for jobs much less repetitive. The idea is simple: → Jobsly analyzes the job description with AI → Understands how well the role matches your profile → Opens the application flow → Automatically fills repetitive fields → Uses AI to generate relevant answers where needed → Helps you move through applications much faster We already have autofill support working for Greenhouse and Workday , and I’m currently working on adding more job providers and improving the autopilot flow. It’s still under development, so this isn’t a launch announcement yet — just sharing what I’m building. 🛠️ Planning to open it to the public soon. 🌐 https://jobsly.pro Would love to hear what feature you’d want in an AI job application copilot. buildinpublic ai jobs automation jobsearch chromeextension saas
LUXE Store
I built a full-stack e-commerce platform using the MERN stack, featuring user authentication, product management, cart & order functionality, an admin dashboard, email verification, payment integration, and AI-powered features for an enhanced shopping experience. This project helped me get much more hands-on experience with building and deploying a complete production-style application while exploring how AI can be integrated into real-world web applications. Tech: React • Node.js • Express • MongoDB • JWT • Stripe • Resend • AI Would love to hear your feedback! Github Repo: https://github.com/WaleedImran2007/Luxe-Store Live Demo: https://luxe.waleedimran.me
GestureFX – AI Hand Gesture Visual Effects
Building GestureFX – an AI-powered hand gesture recognition system using Python, OpenCV, and MediaPipe. The project detects hand gestures through a webcam and triggers real-time superhero-inspired visual effects such as: - Energy Shield - Energy Beam - Energy Blast - Portal Effect Exploring Computer Vision, Human-Computer Interaction, and real-time gesture tracking to create a touchless interactive experience.
AI-Powered Event Registration & Automation Workflow
Building an Automated Event Registration System with n8n I’m building an automated event registration and participant management system for Automation Hub 2026 , using n8n as the core workflow automation platform . The idea came from a very practical problem: managing an event manually can quickly become complicated when you have to deal with registrations, different participation fees, payment verification, participant capacity, ticket generation, QR codes, email notifications, and check-in. Instead of handling these processes manually, I designed a workflow that connects the different services together and allows the system to automatically process a participant from the moment they submit the registration form. The project is essentially an end-to-end event automation pipeline . A participant submits a registration form → the workflow processes the information → calculates the appropriate fee → checks available capacity → stores the participant's information → generates registration details → creates a QR code → updates the participant record → and sends the appropriate email notifications. The goal is to reduce manual work, minimise human error, and create a smoother experience for both the participant and the event organisers. The Problem Event registration can involve a surprising amount of repetitive administrative work. For a physical and virtual event, an organiser may need to: Collect participant information Determine whether the participant is attending physically or virtually Calculate the correct registration fee Track payments Maintain a participant database Monitor the number of physical participants Generate unique registration IDs Generate tickets Send confirmation emails Generate QR codes Keep participant records updated Manage check-in Send additional notifications when payment status changes Doing all of this manually becomes difficult as the number of participants increases. For example, if someone registers for the physical session, the system needs to know whether physical capacity is still available. If the participant pays, their payment status needs to be updated. Once their registration is confirmed, they need to receive their registration details and ticket. Rather than having someone manually perform each of these steps, I wanted the workflow to make the process as automatic as possible. The Solution I used n8n to orchestrate the entire workflow . The workflow connects different services and uses conditional logic to determine what should happen to each registration. The main components include: n8n — workflow orchestration and business logic Online Form — participant registration Google Sheets — participant records and registration database Google Drive — document and ticket storage Gmail — automated participant communication QR Code API — QR-code generation The workflow acts as the central layer connecting these services. Instead of every application operating independently, n8n coordinates the entire process. Registration Workflow The process starts when a participant submits the registration form. The form collects information such as: Full name Email address Phone number Participation type Occupation or role Payment receipt Other relevant registration information Once the form is submitted, the data is sent into the n8n workflow. The workflow then begins processing the registration. 1. Form Submission The first stage is receiving the participant's registration data. The workflow captures the submitted information and prepares it for processing. This creates a single entry point into the automation. Rather than requiring an administrator to manually copy information from the form into a spreadsheet, the data flows directly into the workflow. 2. Fee Calculation The system determines the participant's registration fee based on their selected participation type. For example: Physical participation: ₦10,000 Virtual participation: ₦5,000 This means the participant does not need to rely on an administrator to manually calculate the expected payment. The workflow automatically assigns the appropriate amount to the registration record. This also makes it easier to maintain consistency across registrations. 3. Capacity Management One of the important requirements for the event is that physical participation is limited. The physical session is available to the first 50 registered physical participants . Because of this, the workflow needs to know how many physical participants have already registered. The automation retrieves existing participant records and calculates the current number of physical and virtual registrations. The workflow then determines whether capacity is still available. For example: If the physical participant count is below 50, the registration can proceed. If the physical participant count reaches 50, the workflow can prevent additional physical registrations or direct the participant toward the virtual option. This is one of the parts of the project where automation becomes particularly useful because the capacity calculation happens automatically rather than relying on someone to constantly monitor a spreadsheet. 4. Participant Data Storage After the registration information has been processed, the participant's details are stored in Google Sheets. The spreadsheet acts as the central participant database. The record contains information such as: Registration ID Ticket number Full name Email Phone number Participation type Amount Payment status Registration status Receipt Ticket URL QR code Check-in status Check-in time Creation timestamp Payment notification status This creates a structured record for every participant. It also means the organisers have a central place where they can monitor registrations. 5. Registration ID Generation Each participant needs to be uniquely identifiable. The workflow therefore generates a unique registration ID for the participant. This ID becomes part of the participant's registration record and can be used to reference the participant throughout the rest of the event process. The same concept applies to the ticket number. Rather than manually assigning these values, the workflow handles the generation automatically. 6. Ticket Generation After processing the participant's registration, the workflow generates the participant's registration details and ticket. The ticket can then be stored using Google Drive. This gives the participant a digital representation of their registration. The workflow also keeps the ticket URL in the participant's Google Sheets record. This is important because it creates a connection between the participant's database record and their generated ticket. 7. QR Code Generation Another part of the system is QR-code generation. The QR code can be associated with the participant's registration or ticket information. This creates the foundation for an automated check-in system. Instead of relying solely on manually checking names against a spreadsheet, the participant can present their QR code during the event. The QR code can then be used to identify the participant and retrieve their registration information. This makes the system more scalable and provides a better experience for event staff. 8. Automated Email Notifications One of the major benefits of the workflow is automated communication. Instead of an organiser manually sending confirmation emails to every participant, n8n can send the appropriate email automatically. Depending on the participant's registration and payment status, different communications can be triggered. For example, a participant may receive information about: Their registration Their payment status Their registration ID Their ticket Their QR code Further instructions This ensures that communication is consistent and reduces the amount of manual administrative work required. Payment Tracking Payment management is another important part of the system. The registration record contains a payment status field that can be updated as the participant's payment is verified. For example, a participant may initially have: Payment Status: Pending After payment verification, the record can be updated to: Payment Status: Approved The workflow can monitor changes to participant records and react to those changes. This creates the possibility of automatically sending payment confirmation notifications once a payment has been approved. The important concept here is that the participant record becomes a source of truth for the workflow. When the state of the participant changes, the automation can respond accordingly. Event Check-In The automation also forms the foundation for an event check-in system. A participant's record contains fields such as: Checked In and Check-In Time When a participant arrives, their QR code can be used to identify them. The system can then update their registration record to indicate that they have checked in. For example: Checked In: Yes and Check-In Time: 9:14 AM This provides the organisers with a real-time record of attendance. It also makes it easier to distinguish between registered participants and participants who actually attended the event. Conditional Logic One of the most interesting aspects of this project is the use of conditional logic. The workflow does not simply execute the same steps for every participant. Different conditions can lead to different branches. For example: Is the participant attending physically? If yes, check physical capacity. If no, continue with virtual registration. Another example: Has the payment been approved? If yes, trigger the appropriate confirmation process. If no, keep the registration in the appropriate pending state. This means the workflow behaves more like a business process engine rather than simply being a sequence of API calls. Why n8n? I chose n8n because it provides the flexibility to visually design workflows while still allowing custom logic and integrations. The visual workflow makes it easier to understand the entire process. At the same time, custom JavaScript can be used when the built-in nodes are not enough. For example, I used custom logic for capacity calculations and participant processing. This combination of visual automation and programmable logic makes n8n particularly useful for building practical automation systems. Architecture At a high level, the architecture looks like this: Participant ↓ Registration Form ↓ n8n Workflow ↓ Process Registration ↓ Calculate Fee ↓ Check Capacity ↓ Store Participant Data ↓ Generate Registration Details ↓ Generate Ticket ↓ Generate QR Code ↓ Update Participant Record ↓ Send Email Notification This architecture allows each stage of the process to perform a specific responsibility. It also makes the workflow easier to maintain because individual stages can be modified without completely rebuilding the system. Building for a Real Event What makes this project particularly interesting to me is that it isn't just a demonstration workflow. It is being designed around a real event with real participants . That changes the way I think about automation. When building a tutorial workflow, it is easy to ignore edge cases. With a real event, however, things such as duplicate registrations, capacity limits, payment status, missing information, invalid submissions, and failed notifications become important. The workflow therefore has to be designed with actual operational scenarios in mind. This project has also helped me understand that automation is not simply about connecting tools. It is about understanding the underlying business process and determining where automation can safely take over repetitive tasks. Lessons From Building It One of the biggest lessons I've learned from this project is that good automation starts with process design . Before creating the n8n nodes, I had to understand the entire registration journey. What happens when someone registers? What happens when the physical capacity is full? What happens when payment is pending? What happens when payment is approved? What information should be stored? What should the participant receive? What happens when the participant arrives at the venue? These questions are just as important as the technical implementation. The workflow is only effective when the business logic behind it is clearly defined. Another important lesson was understanding how different systems communicate. The project required connecting forms, spreadsheets, cloud storage, email, APIs, and custom logic. Each service has its own structure and requirements. n8n provides the orchestration layer that allows these systems to work together. What I Want to Improve This is still an evolving project. There are several areas I would like to improve as the system develops. One area is more robust payment verification. Another is improving the QR-code check-in process so that participant attendance can be captured with minimal manual intervention. I would also like to introduce more intelligent automation around participant communication. For example, an AI layer could potentially help classify participant enquiries, answer common registration questions, or assist organisers with analysing registration data. There is also room for better monitoring and error handling. A production automation system should not only work when everything goes correctly. It should also be able to handle failures gracefully. For example, if an email fails to send, the system should be able to record that failure and potentially retry the operation. If an external API is temporarily unavailable, the workflow should not silently lose the participant's information. These are areas I want to continue exploring. The Bigger Idea Although this project was created for an event, the underlying architecture can be applied to many other use cases. The same approach could be adapted for: Conferences Workshops Training programmes Hackathons Community events Webinars Courses Membership registration Appointment systems Customer onboarding The core principle remains the same: Collect data → process data → apply business rules → update systems → trigger actions. That is the power of workflow automation. Final Thoughts Building this system has been a practical exercise in combining automation, APIs, databases, conditional logic, cloud services, and real-world business requirements . What started as a need to simplify event registration became an opportunity to build a complete automated workflow around the participant journey. The project demonstrates how tools like n8n can be used to move beyond simple task automation and build systems that coordinate multiple processes. For me, the most exciting part is not simply that the workflow works. It is the fact that the workflow is solving a real problem. Instead of manually managing every registration, the system can process information automatically, maintain participant records, manage capacity, generate registration assets, communicate with participants, and support the check-in process. This is the direction I want to continue exploring: building practical AI and automation systems that solve real-world problems. And this project is just one step in that journey.
I Built an AI Finance Tracker with WhatsApp & Google Sheets
I built a practical AI-powered finance tracker that makes recording and organizing personal finances easier. The workflow allows me to send income and expense information through WhatsApp, process the data automatically, and organize it in Google Sheets. The goal was simple: turn everyday financial messages into structured records without manually entering everything into a spreadsheet. Built with: • JavaScript • React • Google Sheets • API Integration • Automation • WhatsApp This project is part of my journey building practical software that solves real-world problems. What would you improve or add to this finance tracker?
I Built a Real Business Website with an AI Assistant & Visitor Tracking
I built a real-world business website for Master Baja Bangunan — a building-materials business in Tangerang, Indonesia. But I didn't want to build just another company website. I wanted to turn a traditional building-materials business into a more interactive digital experience. The website includes: 🤖 AI-powered Material Assistant Visitors can ask questions about building materials and get help estimating materials such as cement, sand, lightweight bricks, rebar, and more. 📊 Visitor Tracking The website includes visitor information so the business can understand website activity. 💬 WhatsApp Integration Customers can quickly contact the business directly from the website. 🌐 Responsive Business Website Designed to present products, pricing, location, business information, and customer contact options in one place. 🌍 Bilingual Experience The website supports Indonesian and English. This project taught me something important: A good website isn't just about how it looks. It's about solving a real business problem. I built this project from scratch as a self-directed developer, combining web development, AI integration, automation, and real-world business requirements. 🔗 Live website: https://masterbajabangunan.my.id/ I'd love to hear from other developers: What would you add to this system next? webdevelopment ai automation javascript aiintegration business fullstack portfolio
Spot the Scam — AI Phishing & Scam Awareness Trainer 🛡️
Shipped my AI-powered phishing/scam awareness trainer! 🚀 Paste any suspicious message (email, SMS, WhatsApp) and it flags red flags instantly using AI — plus a quiz mode to test your scam-spotting skills. Combined my SOC/networking background with AI to build something practical, not just another demo. Stack: - Frontend: Plain HTML/CSS/JS - Backend: Vercel serverless function (api/analyze.js) - AI: Groq API (openai/gpt-oss-20b) Also building a companion Chrome extension for on-the-fly message analysis. 🔗 Try it live: https://security-awareness-trainer.vercel.app/ Would love feedback — try pasting a scammy message and see what it catches! 👇 ai security saas
Shipped: AI Code Explainer — paste code, get a plain-language explanation
Built a small tool that explains any code snippet in plain language — useful when you're reading unfamiliar code or just want a quick refresher. What it does: - Explain like I'm a beginner / intermediate / concise / line-by-line modes - Auto-detects the programming language - Saves explanation history locally - Clean dark UI, built from scratch Stack: HTML/CSS/JS frontend, Node.js serverless backend (Vercel), Groq API for the AI part. Live: [https://ai-code-explainer-fj7e3o9fw-qurat-ul-ain1.vercel.app/] Would love any feedback — especially on what explanation modes or features would be useful to add next.
Multi-Agent System built with n8n & LangChain
Multi-Agent AI System with n8n 🚀 Built a modular multi-agent workflow using n8n to handle complex business task routing and automation. Key Features: Supervisor Agent: Central router to delegate tasks to specialized sub-agents. Task-Specific Agents: Dedicated agents for web research, data extraction, and summary generation. Error Handling & Fallbacks: Ensures reliable execution and output validation. Stack: n8n, OpenAI, LangChain framework logic. Would love to hear feedback from the community! ai automation n8n
Hey DevConnect! Building AI Voice Agents & Workflow Automations
Hey everyone! I'm Ali Hassan , a Computer Engineering student at UET Lahore focused on building production-ready AI systems and agentic workflows . I specialize in bridging full-stack web development with deep automation pipelines—moving past basic demos to build systems that handle real operations end-to-end. 🛠️ The Stack Behind My Automations: • AI Systems & Voice: Vapi, ElevenLabs, Gemini, Claude, OpenAI, n8n, Pinecone, Lovable AI, v0, Bolt.new • Languages & Web: Python, C++, SQL, JavaScript, TypeScript, React, Node.js, FastAPI, Tailwind CSS, REST APIs • Integrations & Cloud: Green API, Twilio, Airtable, Docker, Vercel, Database Design, Google Sheets, Notion API 💡 What I'm currently building: 1. Production AI Voice & Support Agents: Conversational pipelines with sub-second latency, automated CRM sync, and dynamic lead scoring. 2. Academic AI App: An intelligent platform leveraging LLMs to optimize study workflows and context-aware Q&A for students. Excited to connect with fellow AI builders, showcase technical builds, and collaborate! 📁 Portfolio: https://www.alihassan-builds.me/
I Built a 3D Particle Universe You Can Control With Your Hands ✋🌌
What if you could control a 3D world without a mouse or keyboard? I built AetherParticles 3D — an interactive particle universe where hand gestures become the controls. Just a camera, your hands, and thousands of particles. ✋ Gesture → Action ✊ Fist → Implode particles 🖐️ Open Palm → Scatter particles 🤏 Pinch → Scale the particle world ☝️ Point → Attract particles like a magnet ✌️ Peace → Switch to the next 3D shape 🤘 Rock → Trigger a particle burst 👍 Thumbs Up → Increase energy 👌 OK → Randomize the experience The particles can transform into different 3D forms such as galaxies, Saturn, hearts, fireworks, lotus shapes, and more. 🪐❤️✨ 🧠 What I Wanted to Explore The goal wasn't simply to create a visually impressive particle system. I wanted to explore a bigger question: What happens when we remove the traditional interface and let humans interact with digital worlds naturally? Instead of clicking a button, moving a mouse, or pressing a key, your hand becomes the interface. ⚙️ What I Built The project combines: Real-time hand tracking Gesture recognition 3D particle simulation Interactive particle physics Dynamic 3D shape transformations Real-time visualization Modern web UI One of the interesting challenges was making the gestures feel responsive without making the particle system unstable or unpredictable. There is still plenty I want to improve: Better gesture accuracy More realistic particle physics Improved performance More interactive shapes More complex gesture combinations Better visual effects 🚀 Try AetherParticles 3D https://normal-amethyst-pnff6pqr.edgeone.dev/ This project started as an experiment with computer vision and 3D graphics. It turned into something I enjoyed much more: Making the human hand the controller of a digital universe. ✋🌌 I'm continuing to learn, build, experiment, and explore where AI + computer vision + creative coding can take interactive experiences next. If you try it, I'd love to know: Which gesture or particle shape should I add next? 👀 showcase ai computervision threejs creativecoding webdev javascript 3d
AI-powered Customer Feedback Sentiment Analyzer & Alert Bot using n8n
An automated AI workflow built in n8n that analyzes customer feedback sentiment in real time, routes actionable insights, and triggers instant alerts for negative feedback! 💡 The Problem Businesses receive feedback across multiple channels, but manually reading, categorizing, and prioritizing urgent complaints takes too much time. Urgent negative feedback often gets lost or responded to late. ✅ What It Does Automated Data Ingestion: Collects customer feedback instantly via Webhooks or Forms. Sentiment & Intent Analysis: Uses AI (Gemini / LLM) to evaluate sentiment (Positive, Neutral, Negative) and extract key topics. Smart Routing & Alerts: Automatically sends real-time alerts (Slack, Telegram, or Email) to support teams when negative sentiment is detected. Data Logging: Logs structured sentiment metrics directly to Google Sheets or Supabase for analytics.
AI Appointment Booking System for HVAC companies
Built an AI Appointment Booking System for HVAC Businesses What if an HVAC company could handle incoming service requests, qualify customers, and book appointments without manually managing every step? I built this workflow using n8n + AI + Calendly/Calendar to automate the process from customer intake to follow-up. The workflow: → Customer submits the service form → AI understands the problem and identifies the issue type → Request is automatically routed to the right service → Calendar availability is checked → Customer gets available appointment options → Booking is created → Confirmation is sent automatically → Follow-up & feedback are handled automatically What this means for an HVAC business: • Less manual work • Faster response to customers • Fewer missed opportunities • Automated appointment scheduling • Structured customer information • Automatic follow-ups • More time for technicians to focus on actual jobs The goal isn't just to build an AI agent. The goal is to build a system that removes repetitive work from the business. I'm currently building more practical AI Agents & n8n automation systems for service businesses. If you're an HVAC company, home-service business, or automation agency looking to automate appointment booking and customer handling, feel free to connect or message me. AI AIAutomation n8n AIagents HVAC Automation AppointmentBooking WorkflowAutomation BusinessAutomation
HVAC Inquiry & Booking & Feedback Automation
What if your appointment process could run itself? I recently built this AI automation workflow for an HVAC business to handle inquiries, bookings, and follow-ups with less manual work. Here’s what happens: → Customer fills out the form → Their details are automatically saved in Google Sheets → Calendly handles the appointment booking → Emergency inquiries are flagged and the inquiry team is notified → The customer receives the necessary email updates → After 24 hours, a feedback message is sent → If the feedback is negative, a follow-up is triggered I wanted to build something practical — not just an AI demo, but a workflow that can actually take repetitive tasks off a team’s plate. Built using: n8n, AI, Calendly, Gmail & Google Sheets. Still learning, building, and improving one automation at a time. If you’re working with a service business that has a lot of repetitive inquiry or booking tasks, feel free to connect with me. AIAutomation n8n WorkflowAutomation HVAC BusinessAutomation AppointmentBooking
AI-Powered Social Media Automation & Management Platform
🚀 Built an AI-powered Social Media Management & Automation Platform using n8n. This project allows users to manage multiple social media accounts from a single frontend and automate content publishing without manually posting on every platform. ✨ What it can do • Connect and manage multiple social media accounts • Publish the same post across multiple platforms • Create dynamic posts with customized content for each platform • Schedule social media posts • Automate recurring content publishing • Manage text, images, and other media • Trigger automated posting workflows through n8n • Monitor and manage publishing workflows from the frontend • Keep social media automation centralized in one dashboard 🤖 Automation The backend is powered by n8n workflows that handle the automation logic, API requests, scheduling, content processing, and publishing. Instead of manually creating and publishing the same content on every platform, the system can take one piece of content and automatically distribute it across connected social media accounts. For dynamic automation, the workflow can generate or transform content before publishing, allowing each platform to receive content optimized for its audience and format. 🏗️ Architecture Frontend → API/Webhooks → n8n → Social Media APIs → Publishing The frontend provides the management interface while n8n acts as the automation engine behind the system. 🎯 Goal The goal of this project is to make social media management faster, more scalable, and less repetitive by combining a modern frontend with powerful workflow automation. automation socialmedia n8n ai marketingautomation
AI-Powered LinkedIn Automation
Built an automated workflow that connects LLM-powered AI generation directly with social execution platforms like LinkedIn. Key Capabilities: • Request Processing: Webhook-driven trigger that parses incoming content prompts and inputs in real time. • Contextual Generation: Integrates LLM nodes with active memory to ensure brand-aligned, high-quality post outputs. • Autonomous Publishing: Automatically formats and schedules content directly to LinkedIn without manual intervention. • Error Handling & Fallbacks: Built-in exception tracking ensures zero data loss during API limits or payload errors. Tech Stack: n8n • LLM / Gemini • Webhooks • LinkedIn API • Google Sheets
Shipped: Nexora - Intelligent AI Web Assistant
Built Nexora, an intelligent AI web assistant designed for seamless, intuitive, and efficient digital interaction. Key Features: Contextual AI capabilities with integrated REST APIs for real-time processing Clean, minimalist UI optimized for smooth human-AI interaction Web search grounding and voice input support Modular Python architecture deployed live via Vercel Tech Stack: Python, REST APIs, Vercel, GitHub Would love to hear your feedback and suggestions from the community!
AI Research Workspace
Most of my earlier projects were built around Express.js for the backend. While working on AI-focused applications, I realized that building serious AI systems requires a different backend ecosystem and a deeper understanding of how AI workflows actually operate. That led me to learn FastAPI, and I used this project to not only learn it, but also polish my backend and AI engineering skills by building something from the ground up. Introducing 🚀 AI Research Workspace (AIRW) — an AI-powered workspace designed to interact with and retrieve information from different types of user data. 🔹 What I explored and implemented: • FastAPI for the backend and API architecture • LangChain & LangGraph for building structured AI workflows and agentic execution • RAG (Retrieval-Augmented Generation) for document-based question answering • Document chunking, embeddings, vector search & hybrid retrieval • Semantic retrieval using embeddings combined with traditional keyword-based retrieval • Query rewriting and retrieval pipelines to improve search quality • MongoDB Atlas Vector Search for storing and retrieving embeddings • Supabase Storage for handling uploaded files and generated content • Groq for LLM-powered generation and reasoning • Tavily for web-based research when web access is enabled • Memory system for retaining useful user-specific information • Tool-based agent execution with planning and execution stages 📚 Supported data & AI capabilities AIRW can work with multiple types of content, including: 📄 PDF documents 📊 CSV & Excel files 🎧 Audio files with transcription and timestamp-based retrieval 🖼️ Images using vision capabilities For documents, the system processes the content into smaller chunks, generates embeddings for those chunks, and stores them for semantic retrieval. When a user asks a question, AIRW retrieves the most relevant chunks. The system can also provide sources, allowing users to see where the information came from 🤖 It isn't limited to document Q&A either. AIRW supports: • Document summarization • Retrieval-based questions • General conversation • Web-assisted research • Memory-based responses • Python-powered analysis and visualization • Excel automation and modification • Multi-step tool execution through an agent workflow One of the things I particularly enjoyed was implementing many of these components myself instead of treating AI frameworks as black boxes. Understanding how chunking, embeddings, retrieval, planning, execution, memory, and tool orchestration work together has been one of the most valuable parts of this project. Github Repo: https://github.com/WaleedImran2007/AI-Research-Workspace Live Demo: https://airw.waleedimran.me Still learning. Still improving. 🚀 AI ArtificialIntelligence AIEngineering GenerativeAI RAG LangChain LangGraph FastAPI Python MongoDB AgenticAI MachineLearning BackendDevelopment FullStackDevelopment
Shipped: Corvit AI Chatbot - Smarter Way to Help Students
What if students could just ask instead of searching 5 different campus pages? I built an AI-powered chatbot for Corvit Systems that answers questions about: • Courses, certifications, and NAVTTC programs • Class timings, instructors, and facilities • Fees, career opportunities, and more Key Features: AI-powered conversational assistance RAG-based knowledge retrieval for accurate answers Context-aware responses + online search Clean web interface Tech Stack: GPT-OSS-120B | RAG | JavaScript | Tailwind CSS | Netlify Goal: Turn a traditional info website into an intelligent student assistant. More improvements coming soon! AI RAG Chatbot WebDev JavaScript EdTech CorvitSystems
E commerce website with great Feature AI virtual on
Building a fashion e commerce website just got a lot easier. This Next.js e commerce template comes with Al features designed specifically for fashion brands. The big one is Al Virtual Try-On, helping shoppers visualize how an outfit could look on them before they buy it. It also includes an Al Product Studio, so brands can create and manage product visuals more efficiently. And it's not just a frontend template. You also get: → Next.js e-commerce frontend → Al Virtual Try-On → Al Product Studio → Full admin dashboard → NestJS backend → Product and store management → A ready-made foundation you can customize and launch from For fashion brands, the virtual try-on is especially interesting because it tackles one of online shopping's biggest problems:
When Great Projects Die Underground
“AI won’t replace you, but someone using AI will.” We hear this statement a lot, especially in tech. But lately, I’ve been looking at it from a slightly different perspective. The conversation around AI often focuses on whether developers will be replaced, but there is another issue we don’t talk about enough: visibility. There are people building useful products, solving real problems and experimenting with technology every day. Yet, somehow, many of these projects never get the attention they deserve. I’ve seen situations where students build solutions to problems around them, only for very few people to know that those solutions even exist. Some are free, some are paid, and some are simply projects created to practice and apply what they’ve learned. Meanwhile, another person who knows nothing about coding come along with a similar idea, package it differently, market it better, build trust around it, and suddenly reach a much larger audience. And that got me thinking. Is building actually enough? For many Computer Science students, building projects is part of the learning process. We learn a programming language, identify something we want to solve, build a project around it, run into problems, debug, research, and sometimes use AI to help us understand or improve what we’ve built. AI doesn’t necessarily remove the learning process. In many cases, it becomes another tool within it. But building something and getting people to use it are two completely different challenges. You can have a technically good product and still have almost no users. You can build something free and useful and still have people paying for another product that solves a similar problem. You can spend weeks learning, designing, coding and debugging, only for your project to remain known by the small circle of people who were around when you built it. And honestly, that part hurts. Maybe the problem isn't always the quality of what we build. Maybe we also need to learn how to communicate what we build. How do we make people aware of it? How do we build trust around it? How do we explain its value to someone who isn't a developer? How do we turn a project from “something I built” into “something people actually want to use”? These are questions I’m still thinking about. Because perhaps the next challenge for many developers isn't just learning how to build. It is learning how to make what they build visible.
HeritageAI
Building HeritageAI 🎙️🤖 — a voice-controlled PowerPoint assistant that lets presenters navigate slides hands-free using natural voice commands. Currently improving it with smarter speech interpretation and AI-powered presentation features. 🚀 🔗 https://bit.ly/4Aj9LP0 AI MachineLearning VoiceAI Python HeritageAI
I quit AI for 7 days.
I quit AI for 7 days. No ChatGPT. No Claude. No Copilot. No AI for coding, studying, writing, research, or brainstorming. And honestly? The hardest part wasn’t solving problems. It was realizing how quickly I had become uncomfortable with solving them myself. On Day 1, I kept reaching for AI whenever I got stuck. By Day 3, something changed. Instead of asking: “What's wrong with this code?” I started asking: “What do I THINK is wrong?” That tiny difference made me slow down, form hypotheses, read documentation, make mistakes, and actually remember what I learned. The biggest lesson? AI wasn’t the problem. My dependency on instant answers was. I came out of those 7 days not wanting to use AI less… I wanted to use it better. Because the goal isn't to stop using AI. The goal is to make sure you can still think when the AI isn't there. I wrote about the full 7-day experiment here: https://medium.com/@hasheramin/i-quit-ai-tools-for-7-days-heres-what-happened-cae2bbf3f3fb?sharedUserId=hasheramin Now I'm curious: If every AI tool disappeared tomorrow, how much of your work could you still do on your own ??? AI ArtificialIntelligence SoftwareEngineering Programming Learning Productivity Developer Technology
Help me get hired pleeeeaaaaaseeeee
I'm currently between jobs. My usual stack is React, Supabase, Google Cloud Platform, FastAPI, and Vercel but I can work with other platforms and frameworks too. Mainly Javascript/Typescript and Python. I am a software engineer but I am leaning towards work with AI integration and AI engineering. Here is my portfolio website: https://kamino.is-a.dev