Data & methodology

Bangladesh has no open, crawlable salary dataset, and Bdjobs.com's robots.txt disallows crawlers. The model is therefore fitted to 110 published salary figures (listed below), not to bulk job posts.

Each estimate combines three things: the job (finance, software, sales...), the kind of employer (multinational FMCG, large private bank, top software company...), and years of experience. Technically: ln(salary) = job + employer + experience + city + noise, fitted by weighted ridge regression with priors. Job circulars and official figures carry the most weight, and single blog estimates the least.

Employer pay levels (× a large local company): Multinational FMCG & tobacco 2.588×; Mobile operator 2.164×; Fintech, mobile money & tech platforms 1.491×; Foreign bank 2.071×; Large private bank / finance company 2.428×; Other private bank 2.016×; Top software company 1.568×; Other software / IT company 0.815×; Other multinational 1.982×; Large local company / group 1×; Garments factory / buying house 0.658×; Small / medium local business 0.585×; Private university 1.042×; NGO (local or international) 0.589×; UN agency (National Officer posts) 4.757×.

Experience. Each employer type has its own pay-growth curve, fitted to figures at different career stages. Bank officer ladders, for example, rise much more slowly than multinational fast tracks: Senior Principal Officers (usually 6-9 years) report a median of about Tk 70,000 across banks. The growth per experience level relative to the shared curve is: Multinational FMCG & tobacco 1.015×; Mobile operator 1.01×; Fintech, mobile money & tech platforms 1.005×; Foreign bank 0.958×; Large private bank / finance company 0.889×; Other private bank 0.759×; Top software company 1.051×; Other software / IT company 1.041×; Other multinational 0.999×; Garments factory / buying house 1.048×; Small / medium local business 1.011×; Private university 1.033×; UN agency (National Officer posts) 1.014×. Estimates are interpolated smoothly between levels, so 2.9 and 3.1 years don't jump, and pay never falls with experience (at least +4% per level).

Changing employer type. If you moved, say from a local company to a bank, years at the same kind of employer count in full. Years elsewhere count 75% if you did a similar job, and 40% if it was a different job. This is an assumption: employers in Bangladesh, especially banks, often give only partial credit for outside experience. The salary check shows the experience it used.

Shared experience curve: entry 1×, junior 1.422×, mid 2.013×, senior 2.855×, lead 3.436×. City adjustments are assumptions: Dhaka area 1×, Chattogram 0.92×, Other city (Sylhet, Khulna, Rajshahi...) 0.82×, District town / rural 0.72×.

Each job-and-employer combination carries a confidence label. Well sourced means two or more figures for that exact combination; rough estimate means the combination borrows the employer's general pay level, so its range is shown wider.

Test: in a leave-one-out test, each figure is predicted from all the others. The median error is 21.8% (mean 25.7%), against 29.2% for the prior assumptions alone. Job circulars are reproduced within about 8.4%.