Djibril – Node.js, React, Typescript

This is a social football (soccer) prediction game. Players predict Premier League fixtures, create or join leagues with friends (or open public leagues), and are scored and ranked on leaderboards after every gameweek. It's built around four game modes, each with its own scoring logic. The app is for football fans who want prediction games with friends "without the old-school admin".

Main features:

• Predictions: master predictions per game mode, with per-league overrides and Last-One-Standing pick/used-team tracking.
• Leagues: private & public leagues, invite codes, public discovery, join-request approval, member roles (owner/admin/member), multi-entry, auto-renew, banning/email-blocking, and per-tier league limits.
• Leaderboards: global per-mode and per-league boards with rank movement and PRO-gated filters (following, start gameweek).
• Automated scoring: a pure scoring engine + scheduled jobs that sync fixtures/results from a football data provider and recompute standings.
• Social: follow users, an activity feed, and public profiles.
• Performance Insights: predictions-vs-reality analytics (PRO), with raw-data export (ELITE).

• Guided the client thought the design of the PRD and project planification.
• Designed and built a full-stack, type-safe monorepo with Turborepo
• Designed a relational data model in PostgreSQL with Drizzle ORM
• Built a deterministic scoring engine for four game modes
• Integrated a third-party football data provider (API-Football) for fixtures, results, head-to-head data, teams and standings.
• Built a scheduled job system (node-cron) to sync results and recompute league standings automatically each gameweek.
• Integrated Stripe end-to-end
• Implemented the full league lifecycle: creation, invite-code/public joining, approval workflows, roles & moderation, competitions, entries and renewals, with subscription-tier limits enforced server-side.

A B2B integration platform that enables energy-tech partners to onboard and scale connected-device products across OEM ecosystems. Built for developers, the platform turns complex OEM integrations into a unified API product with enterprise-grade security, observability, and webhook-driven automation

• Led end-to-end implementation of secure auth and reconnection flows, including MFA token handling, credential lifecycle updates, and protected device registration;
• Delivered customer-facing device APIs (GET/POST patterns for battery/HVAC/solar) with auth guards, validation, pagination/filtering, and clear error handling;
• Improved security and reliability by fixing credential encryption/decryption behavior and tightening environment/live-access gating via customer metadata;
• Built and productionized lifecycle webhook orchestration (Cloud Tasks + Svix patterns), enabling reliable event delivery and reducing integration friction for downstream partners;
• Improved engineering velocity and operational excellence via unified logging, API/dashboard observability, and CI/CD workflow upgrades across the monorepo.

A platform that helps to generate and manage lessons for homeschooled children using AI. It helps create custom lessons, book lists, and projects from the child's interests instantly.

• Designed the project development plan according to the recommendations and the timeline;
• Designed the entire project from scratch as a Solo developer;
• Contributed to shaping the product direction with key insights about AI and best practices;
• Built and admin dashboard as a voluntary work to help management automate many tasks that would otherwise require the intervention of a developer.

An AI executive assistant that manages user inbox by scanning emails and archiving those that don't require the user's attention, writes draft replies using a social graph architecture to make the drafts personalized, prepares users for meetings by conducting deep research about each attendee, joining the meeting with a customized meeting bot and take notes on behalf of the user. The product also sends Podcasts every day to help users catch up on what is happening on the news, and their inbox.

• Built the frontend and backend from scratch including the following:

1. Mobile app;
2. Web app;
3. Desktop app.

• Designed the backend architecture;
• Integrated with different LLM providers;
• Optimized the infrastructure cost;
• Built the meeting recording Bot.

This is a chatbot company that allows users to easily create a custom chatbot on top of powerful LLMs.

• Improved existing LLMs model for better accuracy;
• Conducted research on chunking strategies for embeddings;
• Implemented new features such as Google Drive and notion integration.

An online collaboration tool in the style of notio.so, designed for managing projects, courses, and live team cooperation.

• Implemented front-end components and pages;
• Integrated Agora for video and audio calls;
• Wrote automated tests;
• Connected with the backend API using Apollo GraqhQL;
• Fixed bugs.

A startup based in Silicon Valley that helps companies integrate, organize, and track work and communication spread across apps. It provides a AI agent that perform tasks on behalf of the user.

• Used LangChain and OpenAI to create a Chatbot based on the customer's data (using embeddings);
• Designed the architecture of the backend;
• Deployed the backend on Google App Engine;
• Implemented search with Atlas search;
• Connected the front-end and backend using GraphQL;
• Handled different microservices;
• Created the project from scratch;
• Concerted Figma design to React components;
• Created an atomic design system for the app;
• Built features like Google Docs API, Notion API, and Slack bot integrations.

The product was for developers. It helps us protect our code against man-in-the-middle attacks. While committing code to remote providers like Github, they check the code in order to detect any intrusion using their code scan algorithm inspired by blockchain technology.

• was able to find a solution on how to make sure any commit is being scanned by their algorithm before being pushed to the provider (e.g., Github);
• integrated Github, Gitlab, and Bitbucket APIs to commit code from our backend;
• implemented API endpoints for our custom CLI to fetch lists or repositories, commit code, and authenticate using username, password, OTP, and facial recognition.

A floor management system Tablet application for Sephora.

• Built the web application user interfaces;
• Implemented the backend services and API;
• Integrated the backend;
• Wrote tests.

A telemedicine application. The main aim of the application is to connect patients and doctors around the world. Patients can find doctors, book an appointment, and have a video call session with the doctor. During the session, the patient can wear one of the team's custom medical devices (i.e. cgm diabetes monitor) so that the doctor can see the corresponding information about the patient in order to make better decisions.

• Created the architecture of the project using nx monorepo;
• Made technical decisions for the project and assigned tasks to each member of the team;
• Mentored one junior developer who was on the team;
• Created a project architecture where code (such as type, theme, and models) was shared between all the applications in the monorepo;
• Worked on the front-end, created a styling system (using a combination of emotion and styled system), added components to the storybook, reviewing code.