Work About Contact Blog
Book a 30-min call →
RECRUITMENT & STAFFING

More candidate data should create better recruiter decisions —not more waiting.

Voyantt builds technology across the hiring workflow: search at hundreds-of-millions-of-records scale, document processing, first-round voice screening, assessment, and the multi-tenant products that put those capabilities in front of recruiters and clients.

We automate collection and retrieval. People remain responsible for judgment.

Candidate search matching
hybrid retrieval · rerank · voice screen
<200ms end-to-end search over 330M+ profiles
146 tests passing

Search: relevance, latency, and infrastructure cost belong in the same conversation

Keyword search can miss candidates who describe equivalent experience differently. Pure semantic search can lose important exact signals. At large scale, either approach can become expensive if retrieval is not designed around the workload.

For a staffing platform, we combined dense semantic retrieval and BM42 sparse retrieval in Qdrant, fused the result sets, reranked the strongest candidates, and added DuckDB facets for titles, skills, and locations.

The system searches more than 330M profiles in 50–200ms on a single dedicated node costing roughly $215 per month.

Read: ATS AI Search Engine →

Resume and application intake

Applications arrive as PDF, DOCX, scans, and inconsistent text. We build parsing and extraction pipelines that normalise the information into a structure search and workflow software can use.

Where extraction confidence is low, the product should expose that uncertainty and let a recruiter correct the record instead of quietly turning a model guess into candidate data.

Voice screening: automate repetition, not responsibility

Our voice screening agent initiates a call, follows a consistent question flow, records the conversation, and returns a transcript and audio file for recruiter review. Twilio handles telephony; speech-to-text and GPT manage the conversation; ElevenLabs generates the spoken response.

The agent does not make the hiring decision. Its role is to collect the same initial information consistently and give recruiters a reviewable record.

Read: AI Voice Screening Agent →

Assessment that shows its work

Audio assessment can combine quality screening, transcription, structured scoring, and reasoned output. A trustworthy product needs calibration data, clear rubrics, reviewer controls, and ongoing checks for performance across accents, environments, and roles.

We can build the processing and review system. The client and appropriate domain owners must define and validate the assessment policy.

Multi-tenant staffing products

Recruitment platforms often serve multiple client organisations, employer teams, agencies, and internal roles from one codebase. Tenant isolation must apply through search, records, exports, files, integrations, and administration—not only through the login page.

Our broader SaaS work includes query-level tenant scoping, permission-based access, session controls, queues, portals, payments, and deployment infrastructure.

Where We Contribute

Where we contribute

First Steps

Questions we would ask first

01 Which recruiter activity is high-volume but low-judgment?
02 What makes a search result genuinely relevant for your users?
03 Which candidate data can be processed by external services?
04 Who reviews an automated extraction or assessment?
05 What evidence must be retained for a hiring decision?
06 How are different client organisations kept apart today?
07 What will you measure besides time saved?
Start a conversation

Tell us what you are trying to make work.

No pitch deck. We will ask about the users, the current system, the constraints, and what cannot go wrong. If there is a fit, the next step is a written proposal covering the approach, scope, risk, timeline, and cost.