Product Case Study · AI SaaS
First-round interviews are broken. Smart Hire AI fixed them — entirely.
Kalpita AI Interview combines a live AI avatar, biometric identity verification, real code execution, seven-layer fraud detection, and instant AI scoring — in one browser tab. Result: 80% less recruiter effort and 60–70% faster time-to-hire.
80%
less recruiter effort
60–70%
faster time-to-hire
7
layers of fraud detection
24/7
any role, any time zone
Overview
Six things Smart Hire AI changes about first-round hiring
Zero scheduling.
Zero friction. Candidates interview any hour within a 24-hour window — no panel assembly, no calendar coordination, no availability drop-off
Any role.
Any skillset. The interview brain calibrates to IT, non-IT, banking, healthcare, manufacturing, government, education, and BPO domains
Live avatar.
Real conversation. A fully streaming AI interviewer that listens, responds, probes follow-ups, and manages the entire interview in real time — not a video-analysis tool, not a chatbot
7-layer fraud protection.
Biometrics, liveness detection, gaze tracking, deepfake analysis, voice matching, multi-face detection, and temporal pattern monitoring — running simultaneously and invisibly
Built in 6.
8 months by 4–5 engineers using AI-augmented engineering (GitHub Copilot, Claude Code) — a platform competing with what 20+ person teams build over 2 years
No single competitor does all of this.
video analysis, code testing, scheduling, and proctoring vendors each solve one piece; Smart Hire AI does everything, simultaneously, in one browser session
Summary
The hiring industry has not changed in 50 years. This changes it now.
The hiring industry has not fundamentally changed in 50 years. Smart Hire AI changes it now.
Every day, companies hemorrhage time scheduling interviews with people they’ll reject in the first five minutes. Senior engineers and HR managers sit through repetitive first-round screenings instead of work that moves the company forward. Great candidates accept other offers because scheduling took three weeks.
Smart Hire AI ends this — not by optimizing the old process, but by replacing it. Candidates receive a link and join from their browser: no app, no account, no installation. A conversational AI avatar greets them by name, references their resume, and conducts a calibrated interview — technical questions matched to their claimed stack, behavioral probes using the STAR framework, and live coding challenges executed in a built-in code editor. Seven fraud detection layers run silently throughout. The moment the session ends, an AI-generated scorecard lands on the hiring team’s dashboard.
Product profile
At a glance
Smart Hire AI — sold commercially as Kalpita AI Interview — is a cloud-native SaaS platform that runs the complete first-round interview in the candidate’s browser, with no scheduling and no recruiter in the room.
- Product name
- Smart Hire AI (commercially: Kalpita AI Interview)
- Category
- AI SaaS — Automated Interviewing & Talent Assessment
- Built by
- Kalpita Technologies — 130+ professionals, 8+ years, Bengaluru & Austin
- Development
- 4–5 engineers · 6–8 months · AI-augmented (GitHub Copilot, Claude Code, Multi-LLM Strategy)
- Deployment
- Cloud-native SaaS on Azure · browser-based, zero installation · always on
- Who it's for
- Enterprise hiring teams, staffing agencies, educational institutions, government bodies, certification authorities
- Credentials
- SOC 2 (AICPA) · ISO certified · Great Place to Work · Clutch 4.7 · Glassdoor 4.5 · Google 4.7
Tools & technologies
The stack behind the interviewer
- Backend
- ASP.NET Core 9 (C#) — REST API, SignalR hubs, background services — layered architecture: Controllers → Managers → Repository → DataAccess
- Frontend
- Angular 20, TypeScript 5.9 — Standalone components, signals, lazy-loading, RxJS 7.8, OnPush change detection — performant even in low-bandwidth interview conditions
- Real-Time
- ASP.NET SignalR — Bidirectional streaming for transcripts, fraud alerts, upload progress, and full session lifecycle events
- AI Avatar
- HeyGen Live Avatar API — Streaming conversational avatar — speaks, listens, responds, and receives prompt injections in real time
- LLM — Primary
- Azure OpenAI (GPT-4) — Question generation, dynamic follow-ups, STAR analysis, code evaluation, confidence-adjusted scoring
- LLM — Fallback
- Google Gemini — Automatic fallback when primary is circuit-broken — zero-downtime AI availability
- Speech TTS/STT
- Azure Speech — en-US-AriaNeural — Neural voice synthesis (16 kHz) and real-time speech-to-text with 25 dB silence detection
The challenge
The problem we solved
80% of first-round candidates are rejected before a senior interviewer ever sees them — yet those interviews consume full days of senior engineering time.
80% of first-round candidates are rejected before a senior interviewer ever sees them — yet those interviews consume full panel time, recruiter bandwidth, and calendar space, every working day.
Days to weeks
of scheduling delay before the first candidate is even seen.
Up to 40%
candidate drop-off from scheduling friction alone — the best candidates do not wait three weeks for a screening call.
Zero record
traditional interviews leave no structured data, transcript or audit trail.
Inconsistent
different interviewers, different questions, different standards, different biases.
The solution
A live AI avatar that runs the whole first round.
A live, streaming AI avatar conducts the complete first-round interview in the candidate’s browser:
Calibrated questioning
technical questions matched to the claimed stack, behavioral probes on the STAR framework
Real code execution
a built-in code editor where candidates write and run code while the AI probes their reasoning
Seven-layer fraud detection
biometric identity, liveness, gaze tracking, deepfake analysis, voice matching, multi-face detection, temporal pattern monitoring
Instant AI scorecards
structured evaluation in the dashboard the moment the session ends
24/7 operation
candidates choose their hour inside the window; global teams hire across time zones without coordination
Fraud detection
Seven layers between a candidate and a fake interview
Remote interviewing created a new fraud surface. Smart Hire AI closes it with seven independent checks running throughout the session.
Biometric identity verification
Azure Face API confirms the candidate on camera is the candidate who was invited.
Liveness detection
Distinguishes a live human from a photo, a video replay or a recorded loop.
Eye-gaze tracking
WebGazer.js reads gaze direction from the webcam feed, entirely client-side.
Deepfake analysis
Detects synthetic or manipulated video presented as a live feed.
Voice matching
Confirms the speaking voice belongs to the verified candidate throughout the session.
Multi-face detection
Flags a second person in frame coaching or standing in for the candidate.
Environment validation
A room scan validates the interview environment before the session begins.
Inside Smart Hire AI
Results
What Smart Hire AI delivers, measured
The platform was built by four to five engineers in six to eight months — and competes with products backed by far larger teams.
Recruiter effort
80%
80% reduction
Time-to-hire
60–70%
60–70% faster
Availability
24/7
24/7 — any role, any industry, any scale
Evaluation consistency
Every candidate
Same standards, every candidate, with full audit trail
Conclusion
First-round interviewing can be automated without losing rigor.
Smart Hire AI proves two things at once: that first-round interviewing can be automated without losing rigor — and that AI-augmented engineering changes what small teams can build. Four to five engineers, six to eight months, one platform that replaces an entire screening workflow.
FAQ
Frequently asked questions
See it live
See the interviewer that never sleeps.
Request a live demo of Kalpita AI Interview and watch a full first-round interview run itself — avatar, code execution, fraud checks and scorecard.