Infosys Hiring Process: AI Agent Blueprint for Scale
The Infosys hiring process handles one of the largest fresher pipelines in Indian IT services, with reported headcount at approximately 323,379 as of December 2024 and annual campus intake targets between 15,000 and 20,000.
This blueprint maps where Hyring's AI agents can absorb manual workload across the recruitment funnel, what outputs they produce for hiring teams, and which KPIs keep the model measurable.
This article refers to a planning framework showing where AI can reduce manual effort across the funnel.
The Blueprint
You can tell a hiring model is real when the people running it can point to a number that moved last week. If they cannot, it is a strategy document collecting dust.
This blueprint was built for the Infosys recruitment process running at six-figure application volumes. It contains four components:
- A funnel map built for high-volume hiring workflows.
- A phased deployment plan.
- A KPI scoreboard with risk indicators for each metric.
- An ROI module with adjustable inputs.
Infosys operates campus hiring drives across 100+ engineering colleges per cycle. Candidates pass through aptitude rounds, technical assessments, and HR conversations in very highly compressed timelines.
The operational friction concentrates in the first three stages of that funnel, which is precisely where automation carries the most weight without touching evaluative judgment.
Current State
Infosys closed Q3 FY25 with approximately 323,379 employees, according to its quarterly earnings release filed in January 2025. Trailing twelve-month voluntary attrition stood at roughly 13.7%, down from peaks above 20% during FY23 but still generating continuous replacement demand on top of growth hiring.
Campus intake tells a clear story of scaling up. Infosys onboarded approximately 11,900 fresh graduates in FY24, a sharp 76% correction from the over-hiring peak of ~50,000 in FY23.
Commentary from management circles across recent earnings calls has pointed to sustained fresher recruitment in the 15,000 to 20,000 range for FY25 and FY26 as large deal signings create demand for on-time delivery.
The application-to-offer ratio makes the scale tangible. For every campus position filled, Infosys typically processes ten to twelve applications. A single drive targeting 5,000 hires can pull in 50,000 to 60,000 applications within a two-week window. That volume lands almost entirely on the recruiter team's first-stage workload.
~323379
~11900+
~20k
Pain Points of the Infosys Recruitment Process
In such a volume of applications, a bottleneck in one of these stages will not stay restricted to that stage only. It will spill over to the next stage. A three-day delay in shortlisting will turn into a ten-day delay before a candidate even gets to talk to a human recruiter.
Ten days, in a world where TCS, Wipro, HCLTech, and many more product companies are fishing from the same campus recruitment pool, is long enough to lose a quarter of your shortlisted candidates to competing offers (whether better or not).
L1 phone screening is a 6,000-hour tax on the recruitment team
The first round of phone screening for campus hires at Infosys has a set pattern - educational background checks, basic aptitude questions, questions on the role, and questions on the notice period/availability.
Each call lasts typically around 15-20 minutes. The team has to conduct 20,000 such phone screens in a cycle, which is around 6,600 person-hours of conversations where the questions, the answer keys, the grading logic, and the pass/fail criteria are the same every time.
Engineering panels face a different kind of billable hour loss in the form of unfiltered candidates
If the top of the funnel is slow or unfiltered, the engineering talent gets wasted on interviews that could have been better spent on candidates who could have passed the unfiltered aptitude test.
Infosys sells engineering talent to the market at the cost of billable hours. A lead developer spends five hours a day interviewing candidates in a hiring sprint, which is five hours of unbilled delivery time every day.
This cost does not show up in the recruitment budget but on the project profit and loss, where it may not be monitored unless the delivery manager points it out.
Candidate attrition spikes in the gap between rounds
Campus hires in Indian IT typically hold two to four offers by the time their final semester ends. The Infosys interview process competes on speed as much as brand. A candidate who clears aptitude on Monday and does not receive an interview slot until the following Thursday is actively progressing through other companies’ funnels in between.
This loss in conversion is practically invisible in recruiter dashboards because the candidate simply stops responding rather than formally sending a notice of declination.
Remote integrity checks rely on interviewer alertness, which degrades at volume.
Since 2020, a growing share of L1 and technical interviews at Infosys happen over video. Detecting tab switching, secondary speakers, or identity substitution during a live call depends on the interviewer paying attention to peripheral cues.
By the sixth interview of the day, that attention is diminished due to the limitations of human psychology. Scaled remote hiring without automated proctoring creates a systemic blind spot.
Hyring AI Agents
The stages at which Hyring's agents operate are those at which manual repetition consumes time without requiring any human judgments to be made.
The distinction is clear-cut. If it involves some repeatable rubric with a scored or binary output, it can be handled by the agent, otherwise, it can be handled by a human.
The human-in-the-loop at the final decision stage is essential and non-negotiable.
The AI Resume Screener
Instantly parse and rank resumes based on job-fit signals and technical requirements.
- Rapid high-volume parsing
- Objective skill ranking
- Bias-reduced shortlisting
The AI Phone Screener
Automated L1 verification calls to ensure basic qualifications and interest levels.
- 24/7 automated screening
- Consistent evaluation metrics
- Instant shortlist generation
The AI Coding Interviewer
Technical assessment with live coding environments and real-time logic evaluation.
- Real-time code analysis
- Plagiarism detection
- Automated logical scoring
English Proficiency Test
Standardized evaluation of communication skills, fluency, and professional vocabulary.
- Automated fluency scoring
- Grammar and clarity analysis
- CEFR aligned reporting
Funnel-to-agent
| HIRING STAGE | HYRING PRODUCT | WHAT IT DOES | OUTPUT (WHAT THE TEAM SEES) |
|---|---|---|---|
| SHORTLIST | AI resume screener | Shortlists candidates using resume-based role-fit indicators. | Ranked shortlist + job fit signals. |
| L1 SCREENING | AI phone screener | Runs first-round phone screens and supports shortlisting. | Call summary + score + shortlist. |
| STRUCTURED INTERVIEW | AI video interviewer | Runs structured video interviews and generates candidate reports for review. | Interview process automation + structured scoring. |
| TECH VALIDATION | AI coding interviewer | Conducts coding interviews for engineering roles. | Coding evaluation report. |
| COMMUNICATION | English proficiency test | Adds an English assessment step where roles require it. | Level + skill breakdown. |
HYRING PRODUCT
AI resume screener
WHAT IT DOES
Shortlists candidates using resume-based role-fit indicators.
OUTPUT (WHAT THE TEAM SEES)
Ranked shortlist + job fit signals.
HYRING PRODUCT
AI phone screener
WHAT IT DOES
Runs first-round phone screens and supports shortlisting.
OUTPUT (WHAT THE TEAM SEES)
Call summary + score + shortlist.
HYRING PRODUCT
AI video interviewer
WHAT IT DOES
Runs structured video interviews and generates candidate reports for review.
OUTPUT (WHAT THE TEAM SEES)
Interview process automation + structured scoring.
HYRING PRODUCT
AI coding interviewer
WHAT IT DOES
Conducts coding interviews for engineering roles.
OUTPUT (WHAT THE TEAM SEES)
Coding evaluation report.
HYRING PRODUCT
English proficiency test
WHAT IT DOES
Adds an English assessment step where roles require it.
OUTPUT (WHAT THE TEAM SEES)
Level + skill breakdown.
KPI Scoreboard — Infosys Hiring Model
Time to Screen
Definition: Elapsed time from application receipt to L1 decision.
Risk indicator: Compressing this below 48 hours is definitely a win, but only if the panel-stage rejection rate stays flat. A spike there means the early filter traded speed for accuracy.
Screening Throughput
Definition: Volume of initial screens completed per recruiter per week.
Risk indicator: Throughput climbing while the next-stage advance rate drops means the intake net is too wide. Tighten criteria before scaling volume further.
Interview Completion Rate
Definition: Percentage of candidates who receive an interview invite and finish the full assessment.
Risk indicator: Anything below 60% completion usually points to one of three things: unclear invite instructions, broken access links, or assessment length exceeding 40 minutes.
Early Stage Drop-offs
Definition: The specific funnel step where candidates stop participating.
Risk indicator: Heavy drop-off at the assessment stage (versus the scheduling stage) usually means the test feels disproportionately difficult or long relative to the role level. Campus hires abandon assessments faster than lateral hires.
Panel Hours per Hire
Definition: Total human interviewer hours consumed per finalised hire.
Risk indicator: Track this alongside 90-day retention of new hires. If panel hours drop and retention also drops, the AI filter may be too aggressive on soft-skill signals that panels would have caught.
Pass-through Rate
Definition: Percentage of candidates advancing from one specific stage to the next.
Risk indicator: A 15-point variance in pass-through rates between two campuses recruiting for the same role means the rubric needs campus-level calibration from one to the next, not a single uniform threshold.
Shortlist-to-Offer Ratio
Definition: How many shortlisted candidates does it take to produce one accepted offer?
Risk indicator: This ratio, improving from 8:1 to 4:1 is strong. Improving from 8:1 to 3:2 suggests the model is pruning viable candidates who would have succeeded in the role.
Integrity Flag Rate
Definition: Percentage of AI interviews triggering fraud or cheating alerts.
Risk indicator: A baseline flag rate of 3-5% is normal. A sudden jump to 10%+ from a single campus or geography warrants immediate manual review before drawing conclusions about candidate quality from that source.
Time to First Touch
Definition: Lag between application submission and first outreach to the candidate.
Risk indicator: Automating the first touch is easy. The harder question is whether touches two and three (assessment invite, status update) also arrive on schedule. A fast first touch followed by silence for a week creates worse candidate perception than a consistently moderate pace.
SLA adherence
Definition: Percentage of candidates receiving status updates within X days.
Risk indicator: Rising SLA adherence paired with declining offer acceptance may signal that the updates are too generic. Candidates want progress information, not template confirmations.
Model ROI for High-Volume Scenario
Every number in this section is an illustrative projection. The math is grounded in publicly reported Infosys hiring volumes, but internal cost structures vary. Treat this as a starting framework, not a guarantee.
Annual Savings Potential
$1.96M
in operational and hiring costs (itemized basis).
Reduction in Time-to-Hire
66%
faster from application to accepted offer.
Faster Interview Process
52%
acceleration in scheduling and evaluation throughput.
5-Year ROI Projection
~153%
(Year-1, phased deployment). At full 200,000 annual application volume, savings reach ~$1.96M annually, with payback in under 3 weeks per hiring cycle.
Note: ROI figures below are an illustrative model of potential savings in this specific scenario. Actual results vary by internal workflow costs.
Hours Saved Annually
Three stages account for the majority of manual effort in the Infosys recruitment pipeline - resume screening, L1 phone screens, and early technical panels. This is where the time is spent and how we will recover it.
- Resume screening: 200,000 resumes, average time per resume 8 minutes, total time 26,600 person-hours. Using AI for resume screening, we can reduce this to approximately 2,700 person-hours, which will be the recruiter's time spent on reviewing the shortlists created by the AI.
- L1 phone screens: 40,000 candidates remaining after the shortlisting phase, 20 minutes per candidate, total time 13,300 person-hours. Using AI for phone screening, we will have already completed the calls and will have scored results, leaving us approximately 1,500 person-hours of recruiter time for review.
- Early technical panels: 15,000 candidates remaining after L1, 35 minutes per candidate, 1.5 members on average in each panel, total time 13,100 person-hours. Using AI for the coding and video screening phases, we can reduce the number of candidates reaching this phase by 55%, saving us approximately 7,200 person-hours.
Total estimated hours recovered annually: ~42,900.
Hours by Stage (Manual vs. AI-Assisted)
Source: Hyring Hiring Model
Current State Costing vs. Target State with AI
At a 10:1 application-to-hire ratio, every efficiency gain at the funnel's top reduces cost and effort at every stage below it. Here is what the financial shift looks like at Infosys's reported scale.
| METRIC | CURRENT MANUAL PROCESS | TARGET WITH HYRING AI AGENTS |
|---|---|---|
| L1 Screening Cost | ~$1.6M (recruiter hours at $18/hr loaded, coordinator time, telephony infrastructure) | ~$160K (at ~$1 per AI-conducted screen, plus ~$40K in recruiter review hours) |
| Engineering Panel Load | Engineers conduct 4–5 early technical interviews per day during peak campus hiring. | Panels interview only AI-screened candidates, reducing load to ~2 relevant interviews per day. |
| Interview Scalability | Hiring is limited by interviewer availability, time zones, and scheduling logistics. | Candidates complete AI assessments asynchronously within a scheduled window. |
| Fake Candidate Detection | Candidate evaluation varies due to interviewer fatigue and experience levels. | Automated proctoring monitors identity, voice, screen activity, gaze, and clipboard behavior. |
| Funnel Completion Rate | About 28% of shortlisted candidates reach the offer stage | >72% projected, driven by reduced wait times, self-scheduling, and faster status communication |
Where the $1.96M (itemized basis) Savings Come From
The largest cost recovery does not sit in the recruitment department. It sits in engineering delivery margins and candidate acquisition overhead.
- L1 volume handled by AI agents: At 200,000 applications and a 20-minute average screen, the manual L1 stage alone costs roughly $1.6M in loaded recruiter time. AI phone and video agents conducting 90% of those screens at ~$1 each bring that figure down to approximately $200K. Net recovery: ~$1.4M.
- Engineering hours returned to client delivery: Infosys bills senior engineering talent at rates that vary by practice and geography, but internal estimates for mid-level developers in Indian delivery centres typically range from $35 to $55 per hour loaded. Recovering 7,200 panel hours translates to roughly $250K to $400K in delivery capacity restored. Using a midpoint: ~$325K.
- Scheduling and coordination overhead reduced: Campus drive coordination (slot management, candidate communication, rescheduling loops, no-show follow-ups) absorbs an estimated 8,000 to 10,000 coordinator hours per year at Infosys's scale. Self-scheduling through AI assessment links eliminates approximately 60% of that workload. Estimated recovery: ~$150K.
- Re-screening and integrity investigation costs avoided: Manual fraud detection catches problems late, often after a candidate has progressed two stages. Late-stage disqualification wastes all prior evaluation time. Automated proctoring catching integrity issues at the first AI interview stage avoids an estimated $85K in wasted downstream effort annually.
Total estimated annual savings: ~$1.96M (itemized components above; conservative basis).
Where the Savings Come From
Source: Hyring Hiring Model
Cumulative Savings vs Cost
Platform investment is front-loaded where it is seen that savings accumulate with each hiring cycle. In this model, break-even occurs around the sixth month. That timeline reduces significantly if lateral hiring volumes are layered onto the same Hyring deployment, since the marginal cost per additional AI interview is near-zero once the platform is configured.
What makes the Infosys hiring model particularly responsive to AI deployment is the nature of the first three funnel stages. They are high-volume, rubric-driven, and produce scored outputs.
These are not stages requiring creative evaluation or relationship-based judgement. They require consistency applied at speed, which is exactly where software outperforms manual effort.
Cumulative Savings vs Investment (12-month model)
Month
Source: Hyring Hiring Model
Infosys Hiring Process Plan
A deployment that tries to cover 200,000 applications on day one will break any organisation eventually. Staged rollout protects existing workflows, generates comparison data, and gives hiring panels time to trust the AI outputs before they rely on them.
Week 2
Activate the AI Phone Screener and AI Video Interviewer for the pilot roles only. Run them alongside your existing manual process for the first 72 hours.
Compare outputs - score distributions, time-to-decision, candidate completion rates. Track daily, and not weekly. Surface any discrepancies between AI recommendations and manual decisions immediately so rubrics can be calibrated before volume increases.
Week 3
Add the AI Coding Interviewer for the engineering track. Add the English Proficiency Test for any role where client communication is a first-year requirement. Begin monitoring integrity flag rates.
Compare them against whatever baseline you have from manual fraud detection (even if that baseline is informal). Flag rate spikes from specific campuses or candidate pools should trigger a threshold review, not an automatic candidate rejection. The comparisons should give you a fair idea of what works and what doesn't.
Week 4
Aggregate results across both pilot roles. Key review questions: Did pass-through rates improve without panel satisfaction declining? Did candidate NPS hold steady or improve? Did time-to-offer compress measurably?
Question banks, scoring weights, and flag sensitivity need to be adjusted accordingly based on data. Expand to the next batch of requisitions only after two consecutive review cycles show stable SLAs.
Frequently Asked Questions
How does this hiring model handle large volumes of fresher candidates?
Every candidate in a batch receives their assessment link at once. Hyring AI phone and video agents run thousands of sessions concurrently, so a 50,000-application campus drive clears L1 in days rather than the three to four weeks a manual screening team would need.
What does the first-round automated agent deliver for the Infosys hiring process?
A per-candidate report containing a competency score breakdown, a response summary, flagged areas worth probing further, and an advance-or-hold recommendation. Recruiters review the report in about 90 seconds instead of spending 20 minutes on a phone call collecting the same information.
How does the system prevent interview cheating?
The AI video interviewer tracks signals in parallel — tab switches, secondary faces, voice mismatches, gaze anomalies, lip-sync gaps, and clipboard pastes. These feed into a composite integrity score. No single flag triggers automatic rejection; the team sees which signals fired and decides how to proceed.
Do automated interviews hurt the candidate experience?
They tend to improve it. Candidates pick their own assessment window, skip the scheduling back-and-forth, and receive status updates faster because scoring happens immediately after completion.
Can we customize the coding validation for different roles?
Yes. The AI Coding Interviewer supports Java, Python, C++, JavaScript, SQL, and other language environments with configurable difficulty and problem types per role profile. A backend developer and a data engineer see entirely different assessments without requiring separate platform instances.
How do human technical panels fit into this workflow?
Panels step in only at the final stage. By that point, candidates have cleared resume screening, phone or video assessment, coding validation, and language evaluation. Engineers spend their interview time on problem-solving depth and culture fit rather than re-checking baseline qualifications.