Team Fusion was formed to tackle one of the most challenging problems in software process compliance — automating CMMI audits.
-
Coming together from diverse backgrounds in mobile development, backend engineering, and process quality assurance, the team shares a passion for combining AI and automation to simplify complex manual tasks.
-
The hackathon provided us with the perfect opportunity to combine our expertise and build something truly impactful — AuditMatic, a system that bridges business requirements and engineering deliverables using AI.
+
Team Fusion was formed to tackle one of the most challenging problems in software process compliance — automating CMMI audits. Coming together from diverse backgrounds in mobile development, backend engineering, and process quality assurance, the team shares a passion for combining AI and automation to simplify complex manual tasks. The hackathon provided us with the perfect opportunity to combine our expertise and build something truly impactful — AuditMatic, a system that bridges business requirements and engineering deliverables using AI.
+
> ### **Team Member Roles**
+
>
+
#### 1. Shilpa Gade
+
* **Role: iOS Developer
-
#### 1. **Shilpa Gade**
+
* **Bio: Shilpa is an experienced iOS developer who took the lead on the AuditMatic Frontend. She leveraged her strong mobile UX experience to design and build the intuitive React App, ensuring a high-quality, responsive user interface for displaying complex audit reports.
* **Bio**: Shilpa is an experienced iOS developer. She brings a strong mobile-first perspective to the project, contributing to the overall architecture and product strategy with an eye toward future native application development.
-
* **Specialty**: iOS Development, Swift, Mobile Architecture
+
#### 2. Devang Ghorpade
+
3.
+
* **Role: Backend Developer
-
#### 2. **Devang Ghorpade**
+
* **Bio: Devang is a backend developer specializing in creating scalable APIs. For AuditMatic, he was the architect and implementer of the entire Node.js/Express backend, responsible for the RAG orchestration layer (LangChain), Multi-LLM integration, and setting up the scalable PostgreSQL/pgvector database.
-
* **Role**: Backend Developer
+
* **Speciality: API Architecture, Node.js, Systems Integration, Deep AI/ML System Design (RAG, Vector Databases).
-
* **Bio:** Devang is a backend developer specializing in creating scalable APIs. He is responsible for building the Express.js gateway that handles user requests, manages file uploads, and orchestrates the execution of the Python AI core.
+
#### 3. Melisha Dsouza
-
* **Speciality:** API Architecture, Node.js, Systems Integration
+
* **Role: CMMI Subject Matter Expert (SME)
-
#### **Melisha Dsouza**
+
* **Bio: As a core member of the CMMI team, Melisha provided the essential domain expertise. She was instrumental in defining the compliance criteria, but her crucial project contribution was curating and validating the vast set of sample project documents used to train and test the accuracy of the AI pipeline.
-
* **Role**: CMMI Subject Matter Expert (SME)
+
* **Specialty: CMMI Process, Compliance Auditing, Data Curation and Validation.
-
* **Bio**: As a core member of the CMMI team, Melisha provides the essential domain expertise that drives the project. She is responsible for defining the compliance problems, providing the requirement documents for analysis, and validating that the tool's outputs meet official audit standards.
**1. Phase 0: Ideation and Problem Identification.
+
**
+
- The Spark: Recognized the critical, high-effort pain point of manual CMMI compliance auditing.
+
- The Goal: Define the objective: leverage AI to automate compliance auditing.
+
+
**2. Phase 1: Foundation and Proof of Concept (The Existence Check)
+
**
+
- Focus: Establishing the basic architecture and proving the concept of automation.
-
> ### **Our Hackathon Journey**
+
- What We Did:
+
+
1. Setup the core architecture: React App (Frontend) and Express.js (Node.js) Backend.
+
2. Document Processing: We used the backend to dynamically fetch and scrape data from source documents (no permanent storage of documents).
+
3. LLM Integration: Utilized the Gemini model for document processing. We dynamically created a prompt structure (which was already feeded/defined in a simple data structure) and sent the scraped document data along with this prompt to the Gemini LLM.
+
4. Compliance Check: The LLM's output was used to check for the existence of required documents/headers and basic identifier extraction.
+
+
**3. Phase 2: AI Pivot and Content Validation (Current Focus)
+
+
- Focus: Implementing the RAG pipeline to achieve deep content intelligence.
+
-
+
- What We Did:
-
1. Ideation and planning — defining the CMMI audit problem and designing solution architecture.
-
2. Integrating the AI core with the Express.js gateway and setting up test documents.
-
3. Preparing the demo flow, documentation, and UI presentation.
+
1. Implemented the RAG pipeline (LangChain, pgvector) for semantic search and content validation.
+
2. Integrated the Gemini and OpenAI LLM APIs to act as the core validation engine.
+
3. Established a robust data architecture using PostgreSQL/Sequelize and Redis for state and caching.
+
4. Implemented the Role-Based Access Control (RBAC) system.