A young man finishes a six-month computer course at a training centre. He gets a certificate, a group photograph, and a line in somebody's annual report. Eighteen months later he is at home, watching television, and the certificate is in a plastic folder on top of the almirah. Nobody did anything wrong. The course was well run, the trainers were kind, the attendance was good. And it still failed.
We see this pattern constantly in disability skilling, and it is worth naming plainly: the sector has become very good at running training and very bad at producing employment. Those are not the same activity. Training is something you deliver to a person. Employment is something you negotiate between a person, a family, an employer, a supervisor and a commute. Only one of them changes a household's income.
What follows is a practitioner's account of running vocational training and placement work for young people with intellectual and developmental disabilities. Some of it is uncomfortable for NGOs; some is aimed squarely at HR teams who want to hire but keep stalling at the same three obstacles. The gap between the two is where most of the failure happens.
The pipeline problem: why "training" alone almost always fails
The default model in Indian skilling is course-first. An organisation acquires a lab, a curriculum and a trainer, enrols a batch, certifies them, and then goes looking for jobs. By the time the search starts, the candidate has already been shaped by whatever the centre happened to own.
Reverse the order and everything changes. Start with the employer. Find out what work actually exists within a commutable radius, what a shift looks like, what the entry-level tasks are, what the supervisor's real complaints are. Then design training backwards from that. In our experience, a four-week programme built around a confirmed vacancy outperforms a six-month programme built around a confirmed classroom.
A certificate is a claim about the past. A job offer is a claim about the future. Donors fund the first; families need the second.
This is also why placement rates reported at the end of a batch are close to meaningless. A young person who joined on Monday and was gone by Friday still counts as "placed" in most reporting formats. If your metric can be satisfied by an event that lasts four days, it is not measuring what you think it is measuring.
Our own skilling work under Project Samaan, which trains people with special needs for real work roles, is deliberately organised around employer conversations first — and we would rather report a smaller number that held for a year than a large number that dissolved in a month. Where we can, we publish those figures: [Setu to insert: number of trainees placed in 2025-26, and how many were still employed at 12 months].
Match the job to the ability, not the person to the course
Intellectual and developmental disability is not one profile. Two young men with the same diagnosis and the same certificate can have completely different working capacities — one may have excellent fine motor control and struggle with verbal instruction, the other may follow a five-step verbal sequence perfectly but tire after ninety minutes.
Good job matching starts with a genuine functional assessment rather than a diagnosis. The questions that predict success at work are practical:
- How does this person best receive an instruction — spoken, demonstrated, pictorial, written?
- How long can they sustain focused work before they need a break, and how quickly do they recover?
- What is their tolerance for noise, crowding, heat and interruption?
- Can they travel independently, and if not, what would make that possible?
- What does the person actually want to do? This gets asked far less often than it should.
- What triggers distress, and what is the reliable way to help them settle?
Match against that profile and you get placements that survive. Match against whichever course the centre is running that quarter and you get a data entry operator who cannot read the screen comfortably, or a housekeeping trainee placed in the loudest corner of a food court.
The corollary is that the range of viable roles is much wider than the sector's habitual shortlist. Beyond housekeeping and data entry there is stock replenishment, kitchen prep, laundry, packaging, label checking, document scanning, horticulture, warehouse picking, hospital linen management, café service and light assembly. Many are stable, dignified and genuinely suited to consistent, detail-tolerant workers. Our broader skill and development work and our programme for the transformation of people with special needs both proceed from that wider map of roles.
Job coaches and supported employment: the part nobody funds
Supported employment is a well-established international practice and it is uncomplicated in principle. A job coach accompanies the new employee into the actual workplace, learns the job themselves, breaks it into teachable steps, trains the person on site, and then withdraws gradually as competence builds — while staying reachable if something goes wrong in month four.
This is the single highest-leverage intervention we know of, and it is chronically under-funded because it does not photograph well. There is no batch, no classroom, no graduation ceremony. There is one person standing quietly near a workstation for three weeks.
What the coach does in those weeks decides the outcome:
- Task analysis on site. The real job, in its real sequence, with the real equipment — not a simulated version in a training room.
- Natural supports. Identifying the one colleague on the floor who is patient and willing, and equipping them, so support does not depend on the NGO forever.
- Supervisor translation. Telling the shift in-charge exactly how to give an instruction to this employee, in one sentence they will remember.
- Fade-out planning. Deciding in advance how presence reduces week by week, so the employee does not become dependent on the coach's face being there.
- Crisis protocol. Who calls whom when something goes wrong at 11 a.m. on a Tuesday, before it becomes a termination.
If you are a CSR team reading a proposal, the question worth asking is not "how many will you train" but "how many job coach hours are in this budget, and for how many months after placement". We have written more about how to read a proposal properly in our guide for Noida companies choosing an NGO partner.
Workplace readiness beats technical skill
Ask employers why a placement ended and you will rarely hear that the person could not do the task. You hear about attendance, about a meltdown nobody knew how to handle, about not telling anyone they were unwell, about the bus.
Technical skill is the easy part. The harder curriculum is the one that keeps people employed:
- Travel training on the actual route, at the actual hour, repeated until it is boring.
- Time discipline — arriving, taking a break, returning from a break.
- Asking for help, and telling someone when a task is finished.
- Uniform, grooming and personal care to workplace standard.
- Handling correction from a supervisor without shutting down.
- Basic money handling and understanding a salary slip.
- Knowing what is safe and unsafe conduct at work, and whom to tell.
Every one of these can be taught. Almost none of them are on a standard course syllabus.
Families are part of the placement, whether you plan for it or not
Two family postures end placements early. The first is fear — a mother who has protected her son for twenty-two years and is now being asked to put him on a bus at seven in the morning. The second is a mismatch of expectation — a family that agreed to training in the belief it would end in an office job with a desk, and treats a warehouse role as a humiliation.
Neither is unreasonable. Both need to be worked with before the offer letter, not after. Practical things help: bringing parents to see the workplace, introducing them to the supervisor, being honest about the salary band from day one, and putting them in touch with a family whose adult child has already been working for a year. That last one changes minds faster than any counselling session we have run.
Be candid about money too. Entry wages in these roles are modest. The case for the job is autonomy, structure, identity and a slow upward path — and it is a real case, but only if you make it honestly.
What the law actually gives you: the RPwD Act, 2016
The Rights of Persons with Disabilities Act, 2016 replaced the 1995 legislation and expanded the list of recognised disabilities to twenty-one, including autism spectrum disorder, intellectual disability and specific learning disabilities. It is a more useful instrument than most NGOs and HR teams realise.
The provisions that matter most in employment work:
- Section 34 reserves not less than 4% of vacancies in government establishments for persons with benchmark disability, with 1% specifically earmarked for the category that includes autism, intellectual disability, specific learning disability and mental illness. "Benchmark disability" means 40% or more of a specified disability, certified accordingly.
- Section 20 prohibits discrimination in employment by government establishments and requires reasonable accommodation and appropriate barrier-free conditions.
- Section 21 requires every establishment — not only government ones — to publish an equal opportunity policy. Under the RPwD Rules, 2017, private establishments above the prescribed size must also appoint a liaison officer to oversee recruitment of persons with disabilities, and register the policy with the designated authority.
- Section 22 requires establishments to maintain records of employees with disabilities.
- Section 2(y) defines reasonable accommodation as necessary modification that does not impose a disproportionate burden — the legal hook for task redesign.
Alongside this, the National Trust Act, 1999 remains directly relevant for families of persons with autism, cerebral palsy, intellectual disability and multiple disabilities, particularly for legal guardianship. Families should also be helped to obtain a UDID card, because without certification the statutory route is closed to them.
For scale context: Census 2011 counted 2.68 crore persons with disabilities in India, about 2.21% of the population, and recorded barely a third of them as workers. The National Sample Survey's 76th round in 2018 put disability prevalence at a broadly similar 2.2%. Whatever the precise number, the employment gap is not marginal.
Retention is the only number that means anything
We would retire the placement count tomorrow if the sector let us. Here is what we would replace it with:
- Employed at 90 days, 180 days and 365 days after joining.
- Wage at joining versus wage at twelve months.
- Reason for exit, recorded honestly and categorised.
- Whether the employer hired a second person with a disability afterwards. This is the real test of whether the first placement worked.
- Whether the employee travels to work independently a year on.
If an employer hires one person from you and never calls again, that was not a placement. That was a favour, and favours do not scale.
Exit reasons repay close attention. A cluster of exits at the eight-week mark usually means the job coach faded out too fast. A cluster after a shift change means the roster was altered without anyone thinking about the commute. These are fixable systems problems, and they only become visible if somebody is writing them down.
What employers need to change — and it is not their hiring policy
Most companies we work with already want to hire. The blockage sits further downstream, in three places.
Task redesign, not role creation
Do not invent a job. Take an existing role and unbundle it. In almost any operation there is a set of repetitive, rule-bound tasks currently scattered across three people who each dislike doing them. Consolidate those into one well-defined role and you have created genuine work with genuine productivity, not a sympathy seat. This is what reasonable accommodation looks like in practice: fixed sequence, visual work instructions, a consistent workstation, a predictable shift.
Supervisor sensitisation, not company-wide webinars
The person who decides whether a placement survives is the immediate supervisor, not the CHRO. That supervisor needs perhaps ninety minutes of practical, concrete input: how to give an instruction, how to correct without escalating, what a sensory overload looks like and what to do, whom to call. A one-hour all-staff sensitisation webinar changes nothing on the floor.
Stop hiring as charity
Charity hiring produces a predictable outcome — the employee is given nothing meaningful to do, the team learns that people with disabilities cannot work, and the experiment is quietly ended. Hold the employee to the standard of the role, provide the accommodation the law contemplates, and appraise them like anybody else. Being taken seriously is the accommodation that matters most.
A related point for HR: fix your recruitment process before you fix your headcount target. If the application is an inaccessible PDF, the assessment is a timed aptitude test with dense language, and the interview is unstructured and abstract, you have screened out your candidates before anyone met them. A short work trial tells you more than an interview ever will.
Where to start
If you are an NGO: pick one employer, place one person properly with a real job coach, keep them employed for a year, then go back to that employer for two more. That sequence builds something. A hundred certificates does not.
If you are an HR or CSR team in Noida or the wider NCR: identify one department with unbundleable tasks, brief that one supervisor, and start with two roles rather than twenty. Ask your NGO partner for retention data, not photographs.
And if you are a parent, the earlier functional skills begin, the wider the options at eighteen — a principle that runs through our schooling work too, including our efforts to keep girls in education through Project Uday. To discuss hiring, hosting a work trial, funding job coach hours or referring a young person for assessment, get in touch with our team in Noida.
The measure we hold ourselves to is simple, and we would encourage every organisation in this space to adopt it: not how many people we trained, but how many are still going to work next year.
