An AI-powered knowledge curation and coaching platform that helps users build and maintain expertise through personalized learning paths.

Features an activity recognition system that analyzes user work patterns and maps them to relevant learning content using LLM analysis. Implements a sophisticated lesson management system with automated assignment reviews and progression tracking.

Uses Google OAuth 2.0 authentication integrated with WorkSmart API, and includes content suggestion pipelines that generate insights from meeting transcripts, user activities, and expert content sources. Built on AWS infrastructure with SQS message queuing for scalable processing.

A comprehensive admissions, enrollment, and Student Information System (SIS) for Alpha School and sister schools, managing the complete student lifecycle from initial inquiry through enrollment and rostering.

Built as an integrated ecosystem centered on HubSpot with custom middleware services. Handles lead capture, application processing with Stripe payment integration via custom PayGate service, shadow day scheduling, offer management, and deposit collection with Maxio subscriptions.

Serves multiple Alpha School brands and supports various enrollment types including full-time students, ancillary programs, camps, and marketing events. Integrates with learning platforms like Dash, TimeBack, and CoachBot for post-enrollment student management.

An educational gaming platform that combines engaging Unity gameplay with research-backed learning science principles to build mathematical fluency.

As SVP, I served as both Product Manager and backend developer, designing and implementing the learning algorithms from scratch. The system implements spaced repetition for long-term retention, automaticity building through progressive speed challenges, active learning via continuous problem-solving, and non-interference (guess prevention) to ensure genuine mastery. Tracks individual fact mastery through seven sequential learning stages with dynamic difficulty adjustment.

Features an XP-based motivational model with streak rewards and progress visualization. The game follows a cat navigating obstacles home by solving mathematical puzzles, supporting multiplication, addition, subtraction, and division aligned with Common Core standards. Built on AWS serverless architecture with a few hundred active users.

An educational monitoring and analytics platform deployed at Alpha schools that combines AI-powered surveillance and real-time coaching to enhance student learning.

Built as a desktop Electron application that captures screen content and face/gaze data to create a comprehensive view of student engagement. Supports 50+ learning platforms through a vision-first approach without complex API integrations.

Features include real-time AI analysis and feedback, privacy-conscious design, standardized learning analytics via 1EdTech Caliper, and integration with Alpha's learning ecosystem. The system provides teachers visibility into student engagement that was previously impossible in online learning environments.

An AI-powered comprehensive AP course learning and SAT prep platform designed specifically for homeschoolers, offering personalized learning experiences at approximately $1k per student.

Leverages the CoachBot Platform for core functionality including curriculum, user progress tracking, authentication, and AI-generated content. Features AI-generated video lessons, interactive quizzes, free response questions (FRQ) for skills assessment, and AI tutor support.

Built on AWS serverless architecture using AppSync (GraphQL), Lambda, Cognito, and RDS PostgreSQL with persistent caching. Targets homeschoolers in ESA states with mastery-based progression and adaptive learning powered by CoachBot's algorithms.

A strategic content generation platform that provides the infrastructure for creating, managing, and publishing educational content across Trilogy's learning applications.

As SVP and architect, I designed the platform architecture addressing three core challenges: curriculum modeling (three-level hierarchy from official standards to app-specific curricula with knowledge graph relationships), AI-powered content generation (canonical authoring format optimized for LLMs with programmatic QTI conversion), and safe publishing workflows (draft/production separation with preview capabilities).

The platform serves a 15-person academic team, replacing direct database edits with proper tooling. Features include a Content Generation SDK with shared utilities (data models, QTI converters, LLM tools), a Second Brain knowledge base for pedagogical principles, and standards-compliant delivery via 1EdTech QTI 3.0. Supports both batch generation for reviewed instructional materials and real-time generation for personalized interventions.

A platform that converts and organizes files from Google Drive, Notion, and other sources into a comprehensive, searchable knowledge base for Trilogy's M&A due diligence process.

Leverages AI-powered conversion tools (Claude, OpenAI GPT-4) to transform various file types into searchable Markdown format. Integrates with AWS Bedrock to provide AI-powered question answering over the aggregated knowledge base.

Built on AWS serverless architecture using Lambda, DynamoDB, S3, and Step Functions. Features automated file conversion pipelines, metadata extraction, and centralized knowledge management for analyzing acquired company codebases and documentation.

A mobile application designed to help young children learn to read through personalized, AI-generated content tailored to their reading levels and interests.

Features engaging lessons with tasks, steps, letters, words, and stories. Includes speech synthesis via ElevenlabsAPI and speech recognition capabilities. Uses OpenAI API for evaluating user responses and providing feedback.

Built natively for iOS using SwiftUI, with integration to Apple's learning analytics frameworks. Implements a vision to help students "read what they love, before they love to read" through dynamically generated unlimited reading content that incorporates student interests.

A 1EdTech compliant backend platform focused on OneRoster Rostering and Caliper Analytics standards, providing the infrastructure for TimeBack and other learning applications.

Built on AWS serverless architecture using API Gateway, Lambda, and RDS PostgreSQL. Features comprehensive modules for user management, authentication, event tracking, curriculum management, and analytics integration with QuickSight.

Serves as the central data layer for Alpha's learning ecosystem, handling rostering data and learning analytics for multiple schools and applications.

An educational platform at the core of Trilogy's data-driven, AI-powered approach to highly adaptive, personalized learning. Tracks student activity and progress across various learning apps and generates customized recommendations.

Features comprehensive modules including AI-generated content, adaptive learning recommendations, skills stack tracking, mastery calculations, and quiz generation. Supports multiple learning apps including TeachTower and TeachTap, as well as external platforms like Alpha School.

Implements sophisticated algorithms for personalized learning paths, reward systems, antipattern detection, and analytics integration with AWS QuickSight. Built on AWS serverless infrastructure with extensive curriculum management aligned to educational standards.

Led the engineering team building and operating Crossover's global hiring platform, processing tens of thousands of applications monthly from hundreds of thousands of visitors.

Deeply involved in implementing AI-powered features including automatic AI grading of assessments, AI-based candidate screening and matching, and various AI automations throughout the hiring funnel.

The platform handles the complete hiring lifecycle from sourcing and application processing to assessment management and candidate experience across multiple channels.

An innovative 3D Minecraft-like educational environment designed to gamify the learning experience for students. Visualizes a student's skill stack as interactive blocks in a virtual tower, where each block represents a skill or concept the student has learned or is mastering.

Built with Unity as a multiplayer 3D game integrating with the CoachBot Platform for curriculum management and AI-generated educational content. Featured gameplay modules including combat, pets, avatar customization, multiplayer interactions, and in-app purchases.

The game allowed players to earn diamonds by completing challenges and MCQs, redeem them for real-world prizes, and included premium membership features. Project was discontinued before launch despite significant technical development.

An educational application that allows students in grades 2-6 to practice reading and comprehension skills through highly engaging AI-generated stories tailored to their choices, grade level, and reading proficiency.

Students create and shape their own stories through interactive choices, making reading a fun and magical experience. Features automated content generation with comprehension tracking, progress monitoring, and story sharing capabilities.

Built with React web and mobile apps backed by AWS serverless infrastructure (Lambda, RDS MySQL). Deployed using AWS CDK with automated database migrations and static data seeding.

A self-service subscription management platform enabling Trilogy customers to independently manage renewals, invoices, and account information without sales representative intervention.

Built on top of Salesforce and integrated with Oracle NetSuite/SuiteBilling for billing operations. Features extensive AI automations to streamline subscription renewals and customer interactions, including automated contract renewal workflows.

The system handles both SaaS and non-SaaS products across Trilogy's portfolio, reducing operational costs for low-value ARR subscriptions while improving customer experience.

A "TikTok for education" mobile app designed to help high school students prepare for AP exams and SATs through engaging, personalized, AI-generated content. Featured on Dr. Phil and reached thousands of users before being discontinued due to commercial viability concerns.

Featured AI-animated historical figures as teachers, offering interactive lessons in a scrollable TikTok-style interface with gamified progress tracking. Students could engage in conversations with AI teachers, ask questions, and receive personalized responses.

Built with React Native for cross-platform mobile deployment, integrated with CoachBot Platform for curriculum and progress tracking. Used AWS AppSync (GraphQL), Lambda, RDS MySQL, and Cognito. Notable for its innovative approach to making test prep engaging through AI-powered interactive historical personalities.

I built from scratch a couple of static websites for my wife: a personal travel blog and an interior design presentation site. They are hosted using serverless AWS services.

The content of the sites can be maintained by her using Netlify CMS.

A recently acquired Sales Engagement platform with a micro-service architecture, written in a heterogenous mix of Golang, Rust, NodeJS and PHP, with a React UI and running on AWS.

I have been coordinating multiple parallel tracks including: replacing parts of the system with SaaS offerings, lifting and shifting some of the servers from their data center to AWS, fixing high priority defects, stabilizing and monitoring the product, etc.

Guided a small development team to enhance our customer feedback platform with SSO (Single Sign On) capabilities, by adding OAuth and SAML2 support.

I analyzed the existing code base, derived the architecture, designed the solution and organized the engineering work to ensure the project's success.

I've performed extensive analysis on more than 500 AWS accounts to find cost reduction opportunities.

This involved writing a suite of Lambdas that periodically queried the target accounts for metrics and even executed some strategies (resource deletion, installing agents, etc.).

My main responsibility is to write detailed technical architecture specifications for a portfolio 200+ products. I write about three specs per week, each requiring up to a few hundred man-days to implement.

Most specs extensively use AWS (AppSync, Lambda, DynamoDB, Cognito, etc.) but sometimes we also tackle changes to legacy, monolithic, on-premise products.

We've set up and consumed the AWS Cost and Usage reports to break down and scrutinize the cost structure for an AWS organization generating over 20M USD in yearly costs.

The insights that we found were then implemented as automation and pushed into QuickSight and Google Sheets to be consumed by top executives.

A Spring Boot application with a Semantic UI React frontend for managing which colleagues are coming at the office in which days and in which room / place.

Currently, this product is being sold using a SaaS model, with the option of on-premise deployment on request. The SaaS version runs on AWS (CloudFront, Load Balancer, ECS, RDS).

A YouTube series dedicated to teaching basic web application programming for students.

I am showcasing step-by-step how to build a typical Spring Boot application, backed by an embedded H2 database and fronted by a React user interface.

I was initially the architect and lead developer for building a cloud-based, multi-tenant SaaS platform for managing municipal services. In 2021, I served as an occasional consultant for the team.

We implemented a modular NodeJS core, communicating with several serverless micro-services and a monolithic Angular frontend application.

A micro-service architecture web application built for the Service oriented architecture master course.

It is a re-implementation of the older invoice generator it using Spring, Mithril, NodeJS and Docker.

msg.CareerStart is the de-facto flagship internship programme of the company's local branch. It was primarily based on an idea of mine.

I was the main organizer, a technical mentor and an interviewer for the first edition. We mentored eight students that built a three tier application using ABAP, Spring Boot, and Angular. Afterward, most of these students were offered permanent positions at the company.

A micro-service architecture web application built for the Service oriented architecture master course.

It consisted of several NodeJS micro-services, a Zuul API Gateway, a simplified OAuth 2.0 authorization server and a React user interface.

I've hosted the Software Design laboratory for a group of students at T.U. Cluj Napoca.

I prepared my own learning resources and recorded our sessions in which we explored various architectural styles, design patterns and technologies (mostly Spring and React).

A small integration project on AWS for handling newsletter subscription and un-subscription actions to conform to GDPR.

I designed and fully implemented this solution directly on the customer's AWS account.

A small integration project on AWS for parsing XML exports from a legacy system. It consisted of a micro-service and a Lambda triggered by the S3 events.

I played the role of solution architect, designing and aligning the interfaces of the components, whilst the implementation itself was tackled by other developers.

A small application for submitting feedback after or during the IT Conference on SAP Technologies Cluj.

Later on, I have also hosted a small training about cloud native development using this application as an example.

A simple story-like game for kindergarten children based on the "La Polul Nord" story by Marin Sorescu.

I built it for the Design of interactive software systems master course, the challenge being that the UI must be suitable for usage by four-year-olds.

Co-founder for a Romanian startup, KickIT, targeting the amateur soccer scene.

I was responsible for architecture, design, implementation and operations for the cloud-native solution, consisting of a Spring Boot backend, Lambdas and a React PWA deployed on AWS.

A project for our internal finance department to help them manage travel expense reports, calculate allowances and track business trips for our colleagues.

I acted as technical mentor and product owner for the fairly large team of developers assigned to this project.

SAP API Manager command line interface which can be used to deploy, create and remove API Proxies.

I built this CLI in my spare time to help the development process in SCP projects.

Architect and lead developer for building a suite of reusable micro-services supporting business applications for a market-leading manufacturing company.

I've had a major role in guiding our customer through various architectural decisions, whilst also implementing the most technically challenging components.

A continuation of the "Next Generation Underwriting" project, but focused more towards cloud enablement. The project consists of several smaller prototypes or showcases that aim to be deployed on a cloud platform (e.g. AWS).

I had mostly development-related tasks related to building a MEAN stack solution.

A series of technical trainings primarily for entry level developers and colleagues that wanted to shift technologies. The target technologies are: Spring Boot, NodeJS, React, Angular and AWS.

I was responsible for preparing the training materials, presenting them, doing live demos, regularly checking in with the trainees and finally reviewing their work.

Lightweight open-source UI5 library which enables developers to build user interfaces which adapt at runtime to user roles.

The main goal of this library was to decouple the responsibilities of regular UI5 applications: role-based adaptation should be done declaratively and centrally for each app.

Babel plugin for transpiling inline JSX into UI5 render functions. It enables very easy custom control creation by simplifying the render process.

A simple open-source UI5 library for displaying graph data. I built it using a D3 force graph layout.

A customer add-on to FS-RI (Financial Services - Reinsurance) for the General Reinsurance Company.

The add-on consisted of several UI5 applications which pulled various performance indicators calculated by ABAP CDS views and exposed through SAP Gateway OData services.

An internal project for managing our entry-level job candidates. The project was built using Spring Boot and a PostgreSQL database, with React on the front-end.

My role was that of a technical mentor, organizer and product owner.

Project focused on aiding underwriters in their work through predictive analytics and pre-built business rules.

My role in this project was a combination of a solutions architect, development lead and full stack developer.

We were a relatively small team of five developers and we built a series of Java and UI5 MVPs over the span of a year.

This very website, which is meant to showcase my experience and projects.

Initially built with vanilla UI5, then re-built with custom controls using JSX, finally re-built again using React.

A Java standalone application, with an embedded web server and embedded database. The application aims to help small business to store, manage and use their business data.

I have built this project as an exercise in implementing various mechanisms from scratch (user authentication, session management, scheduling, etc.) to enhance my skills.

A remake of an older project, with MongoDB for storing the document templates, Apache Derby + JDBC for storing the various user, template and field configurations and an UI5 interface. The OpenOffice API was called directly from Java to generate the documents.

A small personal project which realized real-time communication between an ABAP based user interface (classic Dynpro), a UI5 interface hosted on the ABAP system and another UI5 interface hosted on SAP Cloud Platform.

An application which can be used to write imperative Java code by dragging and dropping functional blocks. The generated code can also be displayed inside the application, compiled and run. The front-end is done in UI5 and the back-end is built in Java.

Hospital appointment management cloud application realized with JPA and UI5. Real-time notifications were supported by using WebSocket communication between the clients and the cloud.

A SCP, HANA, UI5 project for the human resources industry. The main goal of the project was to automatically parse and classify candidates based on their CVs.

We had a small development team in which I was the technical lead under the guidance of a senior software architect.

A shipment tracking application targeted at the transport insurance industry branch, consisting of two tracking device prototypes (Android, Arduino) which communicate with cloud services, which store the data, send out notifications and process the collected figures to expose analytical scenarios.

As this was my bachelor degree project, I've built it completely from scratch by myself.

A UI-only application for an online, web-based ticketing system. It has three types of users: customers which may open support tickets, employees which must solve the tickets and managers which can oversee the performance of the employees and must solve any customer complaints.

A UI5 app which uses FabricJS to allow the user to draw by drag-and-dropping shapes and import external pictures. The user can also resize, move and change the colors of the shapes, use undo-redo functionality, save or load drawing files and also to export to PNG.

A small bank application on the cloud. Persistency was done with plain JDBC and the user interface with UI5. This application was later remade by using Hibernate, Spring and AngularJS.

A small bookstore on the cloud. Is very similar to the bank applications, but the persistency was done in XML files (stored in a Cloud Document Store) and the user also has the possibility of generating reports in various formats (TXT, XML, CSV).

Several primary insurance small UI5 applications. All of them had SAP Gateway backend services, which exposed RFCs from standard modules. Most of the applications targeted the transport insurance segment and two of them targeted small goods primary insurance.

Various customer development projects centered on the Life Reinsurance Module.

I was part of the product development team for a few months to get acquainted to the module and the underlying framework. Afterwards, I worked on customization projects for large enterprise customers.

This project was built in my spare time for a land surveyor. Using it, one can keep track of all jobs and related court terms. It offers a calendar for seeing court dates and automatically reads data from the Ministry of Justice Web API and the National Agency for Cadaster website to update the data.

A small customer addition to the Financial Services - Policy Management SAP module.

A mathematical functions graph plotting application entirely written in assembly. The user is able to enter the mathematical formula in the console and then the program plots the function's graph.

This is an application built on top of OpenOffice which can be used to manage invoices for a small company. It allows the user to define multiple profiles, it keeps track of customer data and generates OpenOffice invoices using a user-defined template. Receipts may also be generated from an existing invoice.

A application for land surveyors built on top of OpenOffice, which takes input values from the user and optionally an XML data file and generates documents using user specified templates. The values given by the user or read from the XML are either filled directly in the templates or are used to build tables and hide / show sections.

A program specifically designed to manage dedicated servers for Mount and Blade: Warband. It allowed the simultaneous management of several servers, checking the logs, creating reports, scheduling start, stop or restart actions.

A Mount and Blade: Warband mod which aimed to completely change the game concept without forcing players to download anything. For this project, I was the sole developer. It consisted of modified Python scripts which were run by the server, PHP sites, C / C++ log parsers.