Power Transmission × Havish M Consulting

Read every CV like your best recruiter, in seconds

The power transmission company hires engineers for complex transmission projects across India. HMC built a GenAI screening copilot that matches each CV to the job description, scores it on criteria the hiring manager controls, writes interview questions personal to that candidate, flags what needs checking, and answers questions about the CV in plain English.

ClientPower Transmission FunctionHuman Resources · Talent Acquisition ThemeGenAI CV–JD Matching

The Challenge

Specialist roles, hundreds of CVs, little time

Roles like EHV substation execution or transmission line construction need precise experience. Recruiters were reading every CV by hand, and hiring managers often met candidates without a clear view of fit, gaps or what to probe.

Volume of CVs

Each requisition drew large applicant pools from portals, referrals and agencies, in mixed formats.

Keyword screening misses

Simple keyword filters passed buzzword-heavy CVs and missed strong candidates who described work differently.

Red flags found late

Gaps, short tenures and unsupported claims surfaced in interviews, not before them.

Generic interviews

Panels used the same question list for everyone, so strong claims went untested.

HMC Approach

Explainable scoring first, GenAI on top

Rules the TA team can read and adjust, with GenAI for summaries, questions and conversation.

Select a step to see what happened.

Discover · How good recruiters read a CV

Shadowed recruiters and hiring managers.

Must-haves versus nice-to-haves for key roles.
The signals they trust: evidence, numbers, tenure.
The red flags they look for.

Deliverable: Screening rubric per role family.

Design · A transparent score

Seven criteria, weighted by the hiring manager.

Skills taxonomy for power, projects and HR roles.
Evidence-based matching: a skill counts fully only when shown in work history.
Fair-use rules: no scoring on age, gender, religion or photos.

Deliverable: Scoring model and guardrails.

Build · The copilot

One screen for the recruiter.

CV upload (PDF, Word, text) and JD library.
Score, evidence highlights, career timeline.
Personalised questions, red flags and a CV chatbot, with a large language model for live answers.

Deliverable: Working copilot.

Adopt · Pilot on live requisitions

Human decision, machine preparation.

Recruiters review every shortlist; nothing auto-rejects.
Panels receive the interview kit before the interview.
Feedback tunes the taxonomy and weights.

Deliverable: Pilot playbook.

Scale · More roles, more signals

Extends role by role.

New role families added as taxonomy, not code.
Interview feedback linked back to scores.
ATS integration on the roadmap.

Deliverable: Roll-out plan.

The Solution

A working screening copilot, embedded below

Try it with the five sample candidates, or upload your own CV (PDF, Word or text). It runs in your browser with a rule-based engine; add an OpenAI API key in module 04 for live GenAI summaries, questions and chat.

01 · Screening studio

Match a CV to the JD, with evidence

Pick a role and a candidate, or drop in a CV. Matched skills light up in the CV, the score builds from seven criteria, and the career timeline shows tenures and gaps.

Rule-based

Job description

Candidates

⬆Drop a CV herePDF, DOCX or TXT · stays in your browser

Career timeline

CV with evidence highlightedmust-havegood-to-havenumbers

Match score

Criteria

Sample candidates and CVs are synthetic. Scoring ignores age, gender, religion and photographs by design.

02 · Shortlist board

Set the weights, watch the ranking change

The hiring manager decides what matters. Move a weight and every candidate is re-scored live; the coverage matrix shows exactly which must-haves each candidate evidences.

Skill coverage · JD skills × candidates

Evidenced in work historyListed onlyNot found

03 · Interview kit

Questions written for this candidate, and what to check

Questions quote the candidate's own CV and probe missing must-haves, gaps and short stints. Red flags come with severity and evidence. Copy or download the kit for the panel.

Red flagsRule-based

Strengths to confirm

04 · CV copilot

Ask anything about the candidate

Quick summaries, suggested questions and red flags on demand. Without a key it answers from the rule-based analysis; add your OpenAI API key for live, open-ended answers grounded in the CV and JD.

Live GenAI

Offline

OpenAI API key Model Remember key on this device

Your key is sent only from this browser directly to OpenAI, and is kept in memory unless you tick "remember". CV text is sent to OpenAI only in live mode. Use a key with spending limits.

Impact Dashboard

Recruiters spend time on people, not paperwork

Reading and matching shrank; interview preparation got sharper; hiring managers saw a scored shortlist the same week.

Recruiter hours per requisition

By activity, with cumulative share (Pareto)

Days to a shortlist

Distribution across requisitions, by role family

Figures indicative; validated metrics available on request.

Before vs After

How CVs are screened now

Dimension	Before	After
Reading CVs	Every CV read in full	Summary and evidence in seconds
Matching	Keyword search or gut feel	Evidence-weighted score on seven criteria
Priorities	Implicit, varies by recruiter	Weights set by the hiring manager
Red flags	Found in the interview	Flagged before shortlisting, with evidence
Interview questions	Same list for everyone	Personalised to each CV
Decision	Recruiter alone	Recruiter decides, copilot prepares

Key Takeaways

What any TA team can borrow

Evidence beats keywords

Count a skill fully only when the CV shows it being used; lists alone earn partial credit.

Let the hiring manager own the weights

A transparent score with adjustable weights earns trust that a black box never will.

Prepare the panel, don't replace it

Personalised questions and red flags make interviews sharper; people still make the call.