SShortSingh.
Back to feed

Dev Tutorial: Building a Personalized Learning Path Agent with TypeScript and HazelJS

0
·1 views

A developer tutorial published on DEV Community walks through building a multi-agent AI system in TypeScript using the HazelJS framework to create personalized learning paths. The system addresses common challenges in self-directed learning, including goal definition, skill assessment, resource discovery, and progress tracking. It uses five specialized agents — intake, resource search, curriculum planner, progress tracker, and a coaching orchestrator — each handling a distinct part of the learning workflow. HazelJS features such as the @Agent, @Tool, and @Delegate decorators enable the agents to collaborate while remaining independently maintainable. The tutorial argues that agentic AI is better suited than single-model approaches for personalized education because it allows domain-specific reasoning and continuous adaptation to user progress.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Developer Builds Contradiction-Aware Research Agent on Cognee Knowledge Graph

A developer built ChronoScholar, a research memory agent that ingests arXiv papers into a Cognee knowledge graph and flags when newer papers contradict older stored findings, addressing a key limitation of standard RAG systems. The project was created for the WeMakeDevs x Cognee hackathon using a strict test-driven development approach across 8 components and 60 tests. The system uses Gemini 2.5 Flash to classify relationships between paper pairs and achieved a perfect F1 score of 1.0 on a small 10-pair contradiction detection benchmark, producing 494 entities and 1,025 edges from just 10 ingested papers. Despite comprehensive unit testing, the developer found that all five major integration failures were invisible to the mocked test suite and only surfaced when the real system ran. Key bugs uncovered included a Python str.strip() misuse corrupting JSON output, a silently discarded system prompt in the Groq API call, and greedy template variable substitution causing incomplete prompt rendering.

0
ProgrammingDEV Community ·

Developer Builds Local Reasoning Graph Tool to Stop AI Agents Repeating Rejected Decisions

A developer created NodeDex, an open-source local graph tool designed to track the reasoning history of AI coding agents, including why certain decisions were made or discarded. The tool addresses a specific problem where AI agents confidently re-propose solutions that were already tried and rejected, which the developer describes as a status problem rather than a simple memory recall issue. NodeDex stores decisions along with their rationale and records superseded choices via directional edges, preventing agents from treating outdated conclusions as current. Built on local SQLite under an AGPL license, the project graph never leaves the user's machine and can be tested in 60 seconds via a demo command without an API key. The solo-built tool has been on npm for only three days, with the developer actively seeking real-world feedback on whether agents check dead-end records unprompted.

0
ProgrammingDEV Community ·

Strategy Pattern: How to Replace Messy If-Else Chains in Payment Code

A software developer describes encountering an unwieldy if-else chain in a legacy payment module, where each new payment method required editing a single fragile function and duplicating validation logic. A production incident involving Apple Pay prompted a search for a more maintainable design approach. The Strategy Pattern was adopted, which involves encapsulating each algorithm — such as credit card, PayPal, and Apple Pay processing — into its own interchangeable class. A shared PaymentProcessor context delegates work to whichever strategy is injected, without needing to know the implementation details. This restructuring improved testability, reduced regression risk, and allowed new payment methods to be added without modifying existing code.

0
ProgrammingDEV Community ·

Mobile Engineer Amit Gupta Joins Dev.to to Share AI and App Development Insights

Amit Gupta, a mobile engineer with over 10 years of experience, has published his introductory post on the Dev.to developer community platform. He has worked extensively on Android, iOS, Flutter, and React Native applications, including production-scale SDKs and push notification systems. Gupta is now exploring the intersection of mobile development and artificial intelligence, and is currently learning Python. He plans to share tutorials, case studies, and real-world engineering content focused on Android, Kotlin, and related technologies. His goal is to both contribute to and learn from the Dev.to community.