Open a job board in Dhaka today, and the listings are familiar: accountant, sales executive, garments supervisor. What is conspicuously absent is any serious recruitment for data annotation. This sector generated approximately US$ 6.98 billion globally in 2025 and is projected to reach US$ 44.68 billion by 2035, at a compound annual growth rate of 20.4 per cent.
Fifty-four per cent of enterprises worldwide now depend on annotated data to train their AI systems, and 47 per cent are actively increasing their machine learning investments. The graduates who can meet that demand are graduating from Bangladeshi universities every year and finding slim pickings elsewhere.
The mechanics deserve a brief explanation. Every artificial intelligence system, whether it diagnoses cancer, steers an autonomous vehicle, or moderates social media content, must first be trained on data that humans have labelled with precision.
More than 80 per cent of the engineering labour in global machine learning projects is devoted to data preparation and labelling. Someone must draw a bounding box around a pedestrian in a dashcam image. Someone must classify whether a chest X-ray shows signs of pneumonia. Someone must assess whether a chatbot's response is factually sound or subtly dangerous. That someone, in an ideal world for Bangladesh, should be a qualified Bangladeshi graduate earning a living wage from a laptop.
The medical annotation segment is where the argument becomes most urgent. Bangladesh trains approximately 11,000 doctors annually and enrols tens of thousands more across nursing, pharmacy, and biomedical science programmes.
In early 2024, iMerit launched a dedicated AI-driven radiology annotation tool, reflecting the growing recognition that medical annotation cannot be crowdsourced cheaply. Bangladesh's clinical graduates represent a natural supply.
Beyond medicine, the opportunity maps neatly onto what Bangladesh already produces in abundance. Agriculture students and agronomists can label crop disease imagery, soil degradation maps, and livestock condition footage for precision farming AI.
Linguistics and Bangla-language graduates are well positioned to address the acute shortage of annotated data in low-resource South Asian languages, a gap that major foundation model developers are actively working to close.
A handful of Bangladeshi companies have already demonstrated that the model works. Acme AI, founded in Dhaka in 2020, delivers over 50,000 hours of annotation work per month and is a strategic partner to SuperAnnotate and Alegion, and is one of the youngest companies globally to have won a Grand Challenge by the Bill and Melinda Gates Foundation.
The firm holds ISO 27001, ISO 9001, HIPAA, and GDPR certifications, compliance credentials that provide access to sensitive healthcare and financial datasets. Intellisane AI and Socian Limited, a natural language processing company, have carved out smaller but growing positions in the space. These outfits are, by and large, hiring annotators. The pipeline of graduates to fill those roles is vastly larger than the pipeline of companies organised to absorb them.
The regional comparison is instructive for anyone who doubts the scale of what is being left on the table. The Philippines employs roughly 200,000 workers in data and AI-adjacent roles, drawing on 1.8 million IT-BPM professionals and ranking second in Asia on the EF English Proficiency Index.
India's sector is anchored by domain specialists from engineering and medical colleges, feeding large annotation firms with clients across North America and Europe. Most tellingly, China's National Development and Reform Commission issued guidelines in January 2025 targeting 20 per cent compound growth for the labelling sector by 2027 and creating standardised, nationally recognised AI-training roles. Bangladesh has no equivalent policy architecture.
There are real obstacles that honest observers must acknowledge. Intermittent internet disruptions have affected project delivery timelines for existing clients.
The certification burden for accessing healthcare datasets, including HIPAA and GDPR compliance, requires institutional investment that most small operators cannot afford.
And the absence of a recognised national training standard means that international buyers cannot easily benchmark Bangladeshi annotators against those from the Philippines or Vietnam.
For all that, the broader momentum is shifting in Bangladesh's favour.
The Korea International Cooperation Agency committed US$ 96 million to Bangladesh across five major projects beginning in 2026, including a US$ 13 million initiative to develop a high-tech workforce with a focus on artificial intelligence.
A further US$ 28.5 million project running from 2026 to 2035 will establish an AI Hub Centre at Software Technology Park-2 in Dhaka, with training programmes, doctoral scholarships, and startup incubation among its core components.
The infrastructure is being laid. The question is whether universities, private sector operators, and graduates themselves will organise around it before the opportunity closes.
For a graduate finishing a medical degree or a life sciences programme in Bangladesh today, the arithmetic is worth examining carefully. Remote annotation work for global AI firms offers dollar-denominated income, flexible hours, and a skills ladder that runs from basic image labelling toward machine learning operations and AI quality assurance. Base pricing for many roles at established Bangladeshi annotation firms already starts from US$ 180 per month, with senior and specialist roles commanding significantly more.
Asia-Pacific's data annotation market is forecast to grow at over 25 per cent annually through 2031.
Someone will serve that growth. The graduates of Bangladesh's medical colleges, agricultural universities, and linguistics departments are well placed to serve it. What they are waiting for, and what the country's policymakers and private sector have not yet provided with sufficient urgency, is a coordinated path to do so.
saimanur2003@gmail.com
© 2026 - All Rights with The Financial Express
