OptiBuild is a research-backed construction scheduling and cost-delay optimization engine built for the Savitribai Phule Pune University (SPPU) B.E. Computer Engineering final-year project.
It features a mathematically rigorous Hero Module: CPM Precedence DAG + Resource-Constrained Genetic Algorithm (DEAP) + MahaRERA Section 18 Delay Penalty vs. Crash Cost Optimizer, benchmarked against Google OR-Tools CP-SAT.
To guarantee academic viva defense balance and modular software architecture:
| Member | Module Ownership | Core Tech & Directories | Viva Defense Topic |
|---|---|---|---|
| Divyal Padalkar (Member 1) | CPM + Genetic Algorithm |
networkx, deap (app/engine/cpm/, app/engine/ga/) |
DAG topological traversal, forward/backward passes, GA chromosome encoding & fitness convergence. |
| Chaitanya (Member 2) | CP-SAT Solver & Penalty Engine |
ortools, Set Theory (app/engine/cpsat/, app/engine/penalty/) |
Google OR-Tools exact solver benchmark, optimality gap, MahaRERA Section 18 penalty formula ( |
| Bhushan Bhosale (Member 3) | Full-Stack & UX | FastAPI, Next.js, Jira REST (app/backend/, app/frontend/, jira-feedback/) |
REST API schemas, database migrations, interactive Gantt chart, in-app Jira user feedback webhook. |
| Aniket (Member 4) | Data & Computer Vision |
pdfplumber, MobileNetV2 (app/ingestion/, app/cv/) |
MahaRERA PDF extraction with review-and-confirm UX, binary stage classifier (Structural vs Finishing). |
When team members clone this repository and open it in Antigravity 2.0, all agents automatically share the exact same context, guardrails, and sprint progress:
┌────────────────────────────────────────────────────────────────────────┐
│ Atlassian Jira Cloud │
│ (Sprint backlog, epics, viva burndown) │
└───────────────────▲────────────────────────────────▲───────────────────┘
│ │
Atlassian MCP Tool Atlassian MCP Tool
│ │
┌──────────┴──────────┐ ┌──────────┴──────────┐
│ Antigravity 2.0 (M1)│ │ Antigravity 2.0 (M2)│
└──────────┬──────────┘ └──────────┬──────────┘
│ │
└───────► GitHub Repository ◄───┘
├── AGENTS.md (Auto-loaded rules)
├── ACTIVE_SPRINT.md (Live sync board)
└── .agents/ (Rules, skills, MCP configs)
- Auto-Loaded Master Context (
AGENTS.md): Loaded into every teammate's Antigravity context. Enforces role boundaries, viva scope guardrails, and commit conventions. - Live Team Status Board (
ACTIVE_SPRINT.md): Synchronizes what each teammate's agent is working on across Git pushes/pulls. - Agile Jira Hub (
SCRUM/OPTI): Industry-standard Agile workflow that generates real Sprint Burndown and Velocity charts for Chapter 3/4 of the SPPU project report. - Custom Team Skills (
.agents/skills/):/jira-sync: Pull tickets, transition status, log work./team-sync: Sync local branch withACTIVE_SPRINT.mdand teammate states.
- GitHub Collaborator Invite (Mandatory):
- Even if the repo is public or cloned, push access requires collaborator permissions.
- Ensure the repository owner (
26BB) has added your GitHub username under Repository Settings → Collaborators. - Accept the invitation email or notification from GitHub before pushing any code.
# 1. Clone the repository
git clone https://github.com/26BB/OptiBuild.git
cd OptiBuild
# 2. Create and activate a Python 3.10+ virtual environment
python -m venv venv
# Windows (PowerShell):
.\venv\Scripts\Activate.ps1
# Windows (CMD):
.\venv\Scripts\activate.bat
# Linux/macOS:
source venv/bin/activate
# 3. Install all module dependencies
pip install -r requirements.txt- Open the cloned
OptiBuildfolder in Antigravity 2.0. - Antigravity automatically detects all rules and context.
- Simply introduce yourself naturally in the Antigravity chat:
- "Hi, I am Divyal"
- "Hi, I am Chaitanya"
- "Hi, I am Aniket"
Antigravity automatically recognizes your name, presents the 4 modules so you can pick your preferred role, verifies your Python environment, sets up your feature branch, and starts pair-programming on your module!
OptiBuild/
├── AGENTS.md # Auto-loaded AI context & multi-agent rules
├── ACTIVE_SPRINT.md # Live git-synced team sprint & status board
├── PROJECT_MEMORY.md # Master technical memory & academic roadmap
├── TEAM_ONBOARDING.md # 5-min viva defense & teammate quick-start
├── .agents/
│ ├── rules/ # Antigravity rules (jira-agile-sync, module-contracts)
│ ├── skills/ # Team skills (jira-sync, team-sync)
│ ├── skills.json # Antigravity skill manifest
│ └── mcp_config.json # Atlassian MCP configuration
├── app/ # Source Code
│ ├── engine/ # M1 & M2: CPM, GA, CP-SAT, Penalty
│ ├── backend/ # M3: FastAPI REST APIs
│ ├── frontend/ # M3: Next.js UI Dashboard
│ ├── ingestion/ # M4: MahaRERA PDF Extraction
│ └── cv/ # M4: MobileNetV2 Binary Classifier
├── jira-feedback/ # Live MVP in-app feedback bridge into Jira
├── research/
│ ├── literature-survey/ # 12+ Audited research papers & comparative matrix
│ └── math-model/ # SPPU Set Theory formal specification
└── docs/ # SPPU Synopsis, SRS, Viva Guides