Winbid

Procurement discovery and AI-assisted proposal preparation.

RoleFull Stack Developer

Status: In progress
Platforms: Web, Backend

Technology Stack

React

Express

MongoDB

Ollama

Puppeteer

Socket.IO

Key Challenges

Implementing multi-source RFP aggregation with web scraping from SAM.gov, OpenGov, DemandStar, and Bonfire APIs with different authentication methods and data formats
Coordinating document download, parsing, generation, and export with visible progress and cancellation
Handling complex document parsing for PDFs, DOCX, and Excel files to extract RFP requirements and generate structured proposals
Implementing real-time progress tracking with Socket.io for long-running AI document generation tasks with cancellation support
Creating a dynamic proposal builder with section-by-section AI regeneration, content improvement, and professional DOCX export with custom templates

Key Learnings

Gained experience integrating LLM-backed generation into a staged document-processing workflow
Gained deep experience with government procurement APIs (SAM.gov, OpenGov, DemandStar, Bonfire) including authentication, rate limiting, and data transformation
Learned advanced document processing techniques for parsing and generating professional DOCX documents with custom templates
Implemented sophisticated real-time communication with Socket.io for progress tracking and operation cancellation
Developed expertise in template management systems for different proposal types and client requirements

Overview

Winbid is an AI-assisted proposal preparation platform that combines procurement discovery, document processing, and proposal editing. It brings opportunities from sources including SAM.gov, Bonfire, DemandStar, and OpenGov into a React interface, then connects selected requirements to a staged document-generation workflow. The document service uses Llama through Ollama; separate DeepSeek and Grok integration helpers also exist in the backend.

The workflow connects opportunity discovery, downloaded requirements, generation progress, editable proposal sections, and DOCX export. Socket.IO reports progress and supports cancellation during long-running processing.

Key Features

Staged AI Document Workflow

Document processing separates download, parsing, generation, and final DOCX export. Llama is called through Ollama in the document service; model integration helpers remain separate modules.

Automated RFP Discovery

Aggregates and normalizes RFP data from SAM.gov, Bonfire, DemandStar, and OpenGov with advanced filtering and search capabilities.

Professional DOCX Generation

Creates polished proposal documents with custom templates, company branding, and professional formatting ready for submission.

Real-Time Progress Tracking

WebSocket-based live updates during AI document generation with operation cancellation support and instant UI feedback.

Section-by-Section AI Improvement

Granular control over AI-generated content with the ability to regenerate or improve individual sections based on specific requirements.

Template Management

Flexible template system supporting different proposal types, industries, and client requirements for consistent, professional outputs.

Technical Implementation

The platform is built on a React frontend with a Node.js/Express backend and MongoDB for data persistence. The document service runs a staged processing workflow, calling Llama through Ollama for generation, with separate DeepSeek and Grok helpers in the codebase. Puppeteer handles web scraping for RFP aggregation, while Socket.io enables real-time communication for progress tracking during long-running AI operations.

Model Integration: Llama through Ollama in the document service, with separate Grok and DeepSeek integration helpers
Automated RFP Scraping: Puppeteer-based web scraping from SAM.gov, Bonfire, DemandStar, and OpenGov with data normalization
Real-Time WebSockets: Socket.io for bidirectional communication enabling live progress tracking and operation cancellation
Document Generation: Professional DOCX output with custom templates, company logos, and structured formatting
Template System: Flexible template management for different proposal types and client requirements

External Services and Failure Handling

Procurement requests can fail or take longer than document editing interactions. SAM.gov integration includes request timeouts and retries, while provider-specific scraping and normalization stay behind backend modules. Progress events let the interface distinguish processing stages and give users cancellation controls.

Have a project or opportunity in mind?

Let's discuss employment opportunities or freelance work.

Contact Kunj