SaaS / AI
Smart CRM
A multi-tenant CRM with AI assistance built into the record, summarising history and surfacing the next action for every account.
- Next.js
- TypeScript
- PostgreSQL
- OpenAI
- +1
View Case Study
AI / Web Application
AI Resume Platform
Resume Intelligence & Candidate Matching
A web platform that parses resumes into structured data, matches candidates to roles with an explainable score, and generates tailored documents.
- Next.js
- Python
- FastAPI
- OpenAI
- PostgreSQL
Overview
Project Overview
Recruiters were reading hundreds of inconsistently formatted resumes to fill a handful of roles. The platform parses any document into a structured candidate profile, evaluates fit against a role's stated requirements, and shows exactly which requirement each part of the score came from. Because the scoring is explainable, recruiters can override it — and their overrides feed back into calibration.
Services provided
- AI Automation
- Web Application Development
- Client
- Confidential — specialist recruitment firm
- Industry
- Professional Services
- Timeline
- 11 weeks
- Year
- 2025
- Platforms
- Web application
- Team
- 2 engineers, 1 designer
Challenge
The Challenge
Screening was the bottleneck. Strong candidates were being missed because their wording did not match the job description, and there was no record of why anyone was rejected.
Resumes arrived in every format, including scans, with no consistent structure.
Keyword filters rejected candidates whose equivalent experience was phrased differently.
Screening decisions were undocumented, which was a growing compliance concern.
Producing a tailored candidate summary for a client took a recruiter around 30 minutes.
Solution
What We Built
We made the structured profile the product. Everything downstream — matching, search, document generation — reads from it, and every score is traceable back to the source text.
Built a parsing pipeline with OCR fallback that normalises any input into one profile schema.
Scored candidates per requirement rather than in aggregate, quoting the evidence for each.
Added vector search over experience so equivalent skills surface without exact keyword overlap.
Logged every screening decision, its rationale and its reviewer for auditability.
Generated client-ready summaries from structured data instead of rewriting documents by hand.
Capabilities
Key Features
What AI Resume Platform does in day-to-day use.
Format-agnostic parsing
PDF, DOCX and scanned documents parsed into one structured profile schema.
Explainable matching
Every score breaks down by requirement, with the supporting evidence quoted from the document.
Semantic skill search
Vector search finds equivalent experience even when the wording does not match the job description.
Tailored document generation
Role-specific resume and summary variants produced from the structured profile.
Bias-aware review
Configurable field redaction during first-pass review, with a full audit log of who saw what.
Recruiter feedback loop
Overrides are captured and used to recalibrate scoring against real hiring decisions.
Stack
Technology Stack
The tools this system runs on, grouped by the role they play.
Frontend
- Next.js
- React
- TypeScript
- Tailwind CSS
Backend
- Python
- FastAPI
- Celery
AI
- OpenAI
- Hugging Face
- Vector embeddings
Data
- PostgreSQL
- pgvector
- Object storage
Infrastructure
- Docker
- Vercel
- GitHub Actions
Architecture
How It Works
The path a single request takes through the system, end to end.
UPLOAD01
Document intake
Any format accepted, with OCR fallback for scanned files.
PARSE02
Structure extraction
Content normalised into a validated candidate profile schema.
EMBED03
Index experience
Skills and history embedded for semantic retrieval.
MATCH04
Score against role
Per-requirement scoring with quoted supporting evidence.
OUTPUT05
Review & generate
Recruiter reviews, overrides are logged, documents are generated.
Outcome
Results
What changed for the business after launch.
82%
Screening time reduced
Measured across first-pass review of inbound applications.
30 min → 2 min
Candidate summary generation
Produced from the structured profile rather than written by hand.
100%
Decisions with a recorded rationale
Every screening outcome auditable after the fact.
+27%
Shortlist-to-interview rate
Attributed to semantic matching surfacing overlooked candidates.
Interface
A Closer Look
Screens from the delivered system.
Parsed profile — every field traceable to its source text. 
Match score broken down per requirement with quoted evidence. 
Semantic search across experience rather than keyword matching.
More work
Related Projects
Other systems built on similar foundations.
AI / SaaS
AI Customer Support Platform
A support platform that answers repeat questions from verified sources, drafts replies for agents, and escalates anything it cannot ground in documentation.
- Next.js
- Python
- Claude
- PostgreSQL
- +1
View Case Study
Next step
Want a System Like This?
Tell us what your business needs to automate and we'll map out a practical build.
hello@devrox.comWe reply to every enquiry within one business day.