Open to select freelance projects
Hi, I'm Abhishek Kumar.
I build web applications, reliable backend systems, and AI-powered features that help turn ideas into working products. From a first MVP to improving an existing app, I care about the complete experience—not just getting code to run.
You do not need to figure out every technical detail before reaching out. Start with the problem; we can shape the right solution together.
Practical scope. Clear communication. Code you can continue to own.
Turn a clear idea into a usable first version.
For founders and small teams who need a product experience without stitching together multiple developers.
Make the parts behind your product dependable.
For teams that need a new API, a better data model, reliable integrations or help untangling existing backend code.
Put AI to work on a real business task.
For products that need a focused AI capability—not a chatbot bolted on just to say they use AI.
Improve what you already have before rebuilding it.
For teams with an existing app that needs a feature, third-party integration, bug fix or cleaner handover.
Not sure which service fits? Tell me what you are trying to achieve. We can start there.
Describe your projectA curated mix of product builds, backend systems and experiments. Projects can appear in more than one category; repositories and live demos are linked where available.
FEATUREDA project and task workspace built around team collaboration—not just CRUD screens.
Combines projects and task tracking with role-based access, Server-Sent Events, in-app notifications, real-time chat, message receipts and online presence.
Shows product-dashboard thinking alongside real-time collaboration and multi-user workflows.
FEATUREDPersonal product buildAn AI support assistant businesses can configure, train on their knowledge and embed on a website.
Combines an admin dashboard, protected account flows, business-specific knowledge, Gemini-powered replies and an embeddable chat widget.
Connects an AI capability with a configurable product workflow, knowledge setup and customer-facing widget.
FEATUREDA finance-focused application combining market information, interactive charts and AI-assisted exploration.
Built around a React interface, FastAPI services, MongoDB, JWT authentication, stock/news data workflows and Gemini integration.
Relevant to data-heavy dashboards, external API integration and AI-assisted product features.
FEATUREDBackend projectA role-aware REST API for managing projects, tasks and subtasks with account-verification workflows.
The repository documents access and permissions for Admin, Project Admin and Member roles, JWT access/refresh tokens, secure cookies, input validation, email verification and password-reset flows.
Shows practical access control and account lifecycle flows behind a business application.
FEATUREDA full-stack experiment for generating AI profile avatars and prompt-based video jobs.
The documented flow uses Gemini for profile-image generation, submits video-generation jobs through OpenRouter, stores job metadata in PostgreSQL and checks provider status before presenting results.
Demonstrates third-party AI integrations, multi-step generation workflows and asynchronous job-state handling.
A blogging application that places content management and AI-assisted authoring in the same workflow.
The repository describes public blog discovery, categorized search, admin publishing tools, Gemini-assisted content generation, comment handling, image uploads and dashboard statistics.
Relevant to content workflows, admin tools and AI-assisted productivity features.
An online code-execution prototype that separates API requests from execution work.
Uses a React frontend, Express API, Redis queue, PostgreSQL submission state and a separate worker to run C++, JavaScript and Python submissions, with the UI polling execution status.
A strong example of asynchronous backend design, job state and separation between request handling and execution.
Backend services for a social experience with media, interactions and paginated feeds.
The repository covers JWT authentication, post CRUD, S3 image uploads, pagination, comments and replies, likes, Zod validation and rate limiting.
Relevant to content APIs, file uploads, input validation and request protection.
A community-style developer platform for asking questions, posting answers and voting on content.
The repository describes a Next.js and TypeScript app with authentication, protected routes, a rich-text editor, voting interactions and Appwrite backend services.
Shows a content-driven product with user accounts, interactions and a backend-as-a-service integration.
PrototypeA voting-system prototype focused on role-aware candidate administration and one-vote enforcement.
The repository documents JWT authentication, role-based access, vote restrictions and live count updates. Treat it as a prototype rather than a production election system; use synthetic identifiers for demonstrations.
Demonstrates authorization and enforcing domain rules where duplicate actions must be prevented.
An authentication project focused on protected routes and access-controlled experiences.
The public repository describes JWT-based sessions, bcrypt password hashing, protected routes and a modular Next.js/Node.js application.
Relevant to login flows, protected dashboards and role-aware product access.
Learning buildA learning management backend built while deepening practical NestJS and API design skills.
The repository documents a NestJS and TypeScript API with MongoDB/Mongoose and JWT authentication. It is explicitly a learning build, so it is presented as such rather than as a deployed LMS product.
Shows framework practice around modular backend structure and education-domain resources.
A knowledge workspace for writing notes, generating AI insights and querying saved knowledge conversationally.
The repository describes a Next.js frontend, Node.js/Express backend, MongoDB persistence and OpenRouter-powered AI features, including asynchronous note processing and smart querying.
Relevant to internal knowledge tools, document workflows and AI-assisted information retrieval.
ExperimentAn attendance experiment exploring webcam-based face recognition and timestamped records.
The repository describes Python, OpenCV and a KNN-based recognition workflow that matches faces and records attendance to CSV. It is not presented as an audited identity or access-control system.
Shows exploration outside the main web stack, with an explicit experimental scope.
Learning buildA focused repository for exploring common backend patterns with NestJS.
Covers API structure, authentication, authorization, database access, validation, caching and security topics as a learning reference.
Shows ongoing investment in structured backend engineering.
These are personal projects, learning builds and product experiments—not claimed client case studies. For client work, I share only details that are appropriate to make public.
The details change from project to project, but the way of working should never be mysterious.
We discuss the user, the business goal, current constraints and what a useful first delivery should accomplish.
I turn the brief into a clear scope, milestones, assumptions and deliverables before implementation starts.
You get focused updates, reviewable increments and a chance to give feedback before work drifts off course.
We review the result, document important decisions and cover the setup or deployment steps agreed in the scope.
The roles behind my approach to backend systems, full-stack features and product delivery.
Building and scaling Qedix, an AI-powered pull request review platform for JavaScript and TypeScript repositories, with a focus on evidence-based code analysis, scalable backend systems, and reliable cloud infrastructure.
Worked on full-stack features and backend integrations in a live product environment.
Developed application features across frontend and backend codebases.
Built backend services and REST APIs using Node.js, Express.js and MongoDB.
I choose technologies to fit the problem, the existing codebase and the delivery needs—not the other way around.
Languages and fundamentals I use to implement application logic and work across the stack.
Building responsive, accessible interfaces and dashboard-style product experiences.
Application services, resource-oriented APIs, authentication and integrations.
Patterns and technologies for event delivery, live product updates and communication features.
Data modeling, persistence, caching and work that runs outside the request-response path.
Integrating language models and retrieval workflows into useful product features.
Tools used to develop, test, ship and support web applications.
External services and libraries used across selected projects and experiments.
This is a broad technology map across professional work, personal projects and ongoing exploration. Depth varies by tool and use case; each project card lists its own relevant stack.
A little context on my background, how I think about engineering and what I bring to a project.
Backend · Full-stack · AI integrations
Bengaluru, IndiaI enjoy building things that are useful, reliable, and understandable.
My work sits at the intersection of backend engineering and product development. I spend a lot of time thinking about API boundaries, data models, authentication, integrations and asynchronous jobs—but I also care about whether the person using the product can complete their task without friction.
My professional experience spans backend development, full-stack product work and my current role as a Founding Engineer at Qohere. Alongside that work, I build personal projects to explore product ideas, AI capabilities and the engineering patterns behind them.
When working with a client, I aim to make the technical side feel less opaque: understand the actual need, make the trade-offs clear, build in reviewable steps, and leave behind something that can be maintained.
I am building ByaDev as my personal development brand: a place to show the work I do today, collaborate with clients on useful digital products, and create room for future services as the brand grows. Its guiding idea is simple: Build Your Ambitions. Develop Every Vision.
Cochin University of Science and Technology
B. N. Mandal University, Madhepura
A few things people often want to know before discussing a project.
Web apps and MVPs, API/backend work, AI integrations, workflow automation, and targeted improvements to an existing application are all good starting points. If the scope is unclear, we can begin with a short discovery conversation.
Yes. I can review the current architecture, identify a practical first milestone, and work within the stack where it makes sense instead of proposing a rewrite by default.
After understanding the scope, constraints and priority, I can suggest milestones and a project-specific estimate. I avoid quoting a fixed price before the work is clear.
The goal is a clear handover: readable code, setup guidance and notes about important decisions within the agreed scope. Credentials and deployment ownership should remain with the client wherever practical.
Share the idea, the problem or the part of your app that is slowing you down. A rough brief is completely fine.
No need for a perfect spec. Give me the context and we can work out the next sensible step.