Fragmented records
Educational, medical, therapy, and home observations often live in separate systems and document formats.
← Back to portfolio
Family Technology · AI · Information Management
ParentLensAI
A clearer view of a child's journey.
ParentLensAI is a family-centered platform that brings together educational, medical, therapeutic, and daily-life information in one connected system. It helps parents organize records, explore long-term patterns, and ask questions grounded in their own data.
One dashboard connects records, questions, tasks, observations, and long-term progress.
Why I built it
The challenge was not a lack of information. It was that the information was everywhere.
Families of children with complex needs often manage years of IEPs, evaluations, therapy notes, medical records, medications, school updates, appointments, and daily observations.
I saw this challenge in my own family while helping care for my younger brother, Mason. Important details existed across folders, documents, conversations, and memories, but there was no simple way to see how they connected.
ParentLensAI began as an attempt to turn that fragmented information into a clearer and more useful picture.
The problem
Families need more than file storage.
A folder can preserve information, but it does not explain how one record relates to another, reveal changes over time, or help a parent prepare the right question.
Information overload
Long reports contain valuable details, but locating one specific recommendation or historical change can take significant time.
Lost context
Notes, tasks, medication changes, and school performance may only become meaningful when viewed together.
Difficult preparation
Parents often need to reconstruct months or years of information before school meetings, appointments, or care decisions.
What it does
From scattered records to connected understanding
ParentLensAI combines document intelligence, conversational search, structured tracking, and day-to-day organization in one platform.
Connected document intelligence
Parents can upload IEPs, evaluations, medical records, therapy notes, and other documents. The platform extracts and indexes their content so information can be searched and referenced across files.
Context-aware AI companion
The AI companion combines relevant documents, notes, tasks, trends, and recent conversation history to answer questions with information grounded in the family’s own records.
Progress and trend tracking
Medication, academic, and standardized-testing data can be recorded, filtered, visualized, and reviewed over time to help families recognize patterns and prepare for conversations with professionals.
Daily organization
A central dashboard brings together child information, upcoming tasks, quick notes, document summaries, and recent trends so families can manage both long-term records and everyday details.
How it works
A local-first retrieval and reasoning pipeline
The system transforms family records into searchable information, retrieves the most relevant evidence for each question, and passes that evidence to the selected language model.
Ingest records
PDF, DOCX, TXT, and Markdown documents are uploaded, parsed, divided into overlapping text chunks, and checked for duplicates.
Create searchable representations
A local sentence-transformer model converts each chunk into an embedding stored in a persistent ChromaDB collection.
Retrieve relevant context
Each question is compared with indexed documents, notes, and tasks to identify the most relevant supporting information.
Generate a grounded response
The retrieved context and recent conversation history are organized into a protected prompt and sent to either a local Ollama model or an optional OpenAI model.
Communicating the system
Building the platform also meant learning to explain how its parts work together.
I presented ParentLensAI as an end-to-end system rather than simply an AI chatbot. The architecture connects document processing, vector retrieval, structured records, local and cloud language models, dashboards, and privacy-conscious data handling.
Explaining the architecture helped me evaluate the platform from both a technical and user perspective: what information enters the system, how it is retrieved, what the model receives, and where human judgment still matters.
Inside the platform
Designed for both long-term understanding and everyday use
Privacy and responsible design
Sensitive family information requires a different design philosophy.
ParentLensAI was designed with a local-first architecture. Family documents, embeddings, notes, tasks, profile information, and structured records can remain on the user's own computer.
A local Ollama model allows questions to be answered without sending family records to a cloud language model. When the optional OpenAI mode is selected, only the limited text context required for the current question is sent.
The system also treats retrieved documents as reference data rather than trusted instructions, helping reduce the risk that instructions embedded inside uploaded files can alter the assistant's behavior.
What I learned
Building ParentLensAI taught me that a useful AI system is not just a language model. It requires thoughtful information architecture, careful retrieval, privacy decisions, usable interfaces, and a clear understanding of the people the system is meant to support.
Explore the project
See how ParentLensAI was designed and built.
The GitHub repository includes the application structure, document processing pipeline, retrieval system, dashboard, trend tracking, and language-model integration.
Source Code and Documentation
GitHub Repository
Explore the complete project ↗Project Collection
More Projects
Explore research, software, and engineering →PythonStreamlitRetrieval-Augmented GenerationChromaDBSentence TransformersOllamaOpenAI APIPandasAltair