Revamping Resource Management with AI-Powered MatchSmart

Client Overview

When organizations around the world struggle to match the right talent to the right project at the right time, it’s not just an HR issue — it becomes a strategic bottleneck. At Sigma Infosolutions, we faced this exact challenge. But rather than searching externally for a patchwork solution, we built our own: MatchSmart, an AI-powered platform designed to bring clarity, speed, and intelligence to complex talent and resource planning.

What started as a solution for Sigma quickly proved to be a model for solving a global workforce allocation challenge across industries, geographies, and business sizes.

Project Brief

To tackle these inefficiencies in talent allocation and project planning, Sigma Infosolutions developed MatchSmart — an AI-powered platform designed to smartly match the right people to the right projects using structured data and generative AI. The platform consolidates and cleans fragmented information from resumes, vendor databases, and historical project records to support real-time decision-making.

Business Needs And Challenges

Like trying to find the perfect puzzle piece in a messy drawer, our talent data was all over the place:

Resumes were all shapes and sizes, often uploaded without structure— making it hard for machines (and people) to make sense of them.
Our bench and vendor reports were fragile — a misplaced space or capital letter could break the entire flow and create duplication in the vector database.
Chatting with Claude AI came with limits — token ceilings and session timeouts meant we had to be frugal with how we talked to our AI.

Supporting two minds in one body — SkillMatcher and ProjectMatcher — meant our chat interface needed to remember who it was talking to and why, without confusion.
And syncing documents from OneDrive was a dance of precision — we had to split, summarize, and reassemble PDFs across pages without missing a beat.

These weren’t just technical hiccups — they were real roadblocks slowing down our momentum. We needed a platform that could manage all of this without tripping over its own shoelaces.

Solutions

We rolled up our sleeves and built MatchSmart with a clear goal: let AI do the heavy lifting so our teams can focus on what they do best — delivering great work.

Making Sense of Resume Chaos – We set up an AI-powered parsing pipeline using Claude AI to extract structured, clean metadata from PDFs — think skills, domain experience, and project history, all converted to tidy JSON format. We also validated data formats before inserting anything into our vector database, so nothing slipped through the cracks.
Bench & Vendor Data: Tightrope-Walk Precision – To preserve formatting, we moved to a “download-then-upload” workflow. Bench reports were processed with Pandas, catching case and spacing issues. And Claude was trained to roll with the punches on vendor sheets — even the messy ones.
Two Modes, One Brain – Once a user picked either SkillMatcher or ProjectMatcher, we locked it in. This ensured focused logic and avoided “split personality” errors in the backend. A “Start New Chat” button lets users reset and shift gears if needed.

Smarter Conversations with Token-Savvy AI – We designed dual session tracking:
- Time-based auto-expiry (with a 9-minute heads-up),
  And a token-based hard stop at 35,000 tokens.
The goal? Minimize cost, maximize clarity, and never let a chat session wander off a cliff.
Bulletproof Document Ingestion – From OneDrive, our sync system would fetch PDFs, slice them into 2-page chunks, summarize each chunk via Claude, then pass them through a secondary summarization pass for cohesion. No duplicates, no missed updates — just intelligent document knowledge, searchable and reusable.

Benefits

What started as a fix turned into a strategic asset for Sigma’s operations:

Centralized Resource Intelligence: With a bird’s-eye view of skillsets, vendor options, and past project DNA, decision-makers can now act faster and smarter — no more digging through spreadsheets.
Real-Time Talent Allocation: Need a Python pro with fintech chops for a 6-week sprint? MatchSmart can spot them in seconds, based on up-to-date availability and project alignment. It’s like having an AI staffing analyst on demand.
Faster, Smarter Project Planning: Looking to build a new solution? Pull in past project insights, explore tried-and-tested strategies, and let historical wisdom guide today’s innovation.

Operational Efficiency: Manual resume parsing, bench updates, and vendor checks — once time-draining chores — are now automated and reliable, freeing up our teams for meaningful work
Knowledge Reuse, Activated: MatchSmart doesn’t just store documents — it distills, summarizes, and makes them searchable, turning experience into future advantage.
Laying the Groundwork for Enterprise AI: By integrating Claude AI, sentence transformers, and QDrant, we didn’t just solve a staffing challenge — we built a scalable foundation for generative AI adoption across our business.

Tools & Technologies

Frontend: HTML, CSS, JavaScript	Icons: FontAwesome
Backend: Python with Flask	AI Engine: Anthropic Claude API
Embeddings: Sentence Transformers (BAAI/bge-base-en-v1.5)	Vector Database: QDrant (Cloud + API)
Docs Integration: OneDrive API

Get in touch with us, let’s create value together.

Connect with us