danial@haifa:~$
feh ~/photos/danial.jpg
Md. Danial Dirar at a café table against a brick wall, beneath a print of a great wave and a red sun

~/portfolio

Danial Dirar

I build machine-learning systems and private, local AI for businesses — and study the groundwater under Bangladesh's Barind Tract.

GitHubGoogle ScholarProjectsLinkedIn
danial@haifa: ~
$ whoami
Md. Danial Dirar
ML & Software Engineer, Dhaka
$ cat focus.txt
Machine learning and generative AI
Local AI, tailored for business
Groundwater forecasting, Barind
$ ls research/
mar-2025.pdf mar-2026.pdf thesis/
$ ./contact --now
→ danieldirar@protonmail.com
$

$ cat research/field.log

research

The Barind Tract in northwest Bangladesh irrigates almost all of its dry-season rice from groundwater, and its water table falls by up to 0.39 m a year. My research asks where that water can be put back — managed aquifer recharge through canals and reservoirs — and how far ahead the wells can be forecast.

$ water-budget --tract barind

runoff
50.61 Mm³/yr
recharge
27.10 Mm³/yr
53.5%
sites
3,315
MAR points on 377 km of canals and 257 acres of beels
water table
+2.01 m
at the pilot system, 2015–2021
forecast
109 wells
monthly, in the thesis

# Hossain et al., Applied Water Science 16 (2026) and 15 (2025)

$ ls sites/

godagari/ mohanpur/ niamatpur/ ~/

# hover or tap a site to read its file

Barind Tract (approximate outline)Padma, Jamuna and Teesta, flowing to the Bay of Bengal

$ ls projects/ --featured

all projects
Women carrying sacks on their heads walk between green Boro rice fields in Naogaon, in the Barind Tract, with goats grazing and trees fading into the morning mistwater

Groundwater Level Prediction

2026

Undergraduate thesis at Brac University: monthly groundwater forecasts for 109 wells across Bangladesh's Barind Tract, checked against an independent monitoring network.

PythonPyTorchXGBoostscikit-learn
ml
Haifa HiveMind — the honeycomb mark on the Odyssus theme

Haifa HiveMind

2026

A private, local AI research assistant — drop in your papers, ask grounded questions with citations, and it learns from your feedback. Runs entirely offline on your own machine.

PythonFastAPIReactElectron
ai source live
An RTX 3090 board beside the words 3090 48-GB GPU Validator, with icons for VRAM, memory-stability and validation checks

RTX 3090 48GB Mod Validator

2026

A workshop acceptance kit for RTX 3090 cards modded from 24 GB to 48 GB — it checks the card end to end and returns ACCEPT, REVIEW or REJECT before handover.

PythonPyTorchCUDAnvidia-smi
systems source
A Home Inn render of a double-height living room: a walnut media wall, a long sofa and floor-to-ceiling glass beside a brick screen

Home Inn Interior Solution

2026

The first website for a Dhaka interior-design firm with 73 corporate projects behind it — a marketing site, enquiry pipeline and admin panel in one monorepo.

Next.jsNestJSPrismaPostgreSQL
webclient work source
Global River Database: the dashboard with sections for river records, research data, comments and user management, beside the line 'River data, organized.'water

GRD — Global River Database

2025

A shared database of the world's rivers, with the geographical, environmental and financial data researchers need beside each one.

Next.jsNestJSTypeORMMySQL
web source
CreditSense AI — a gold classical bank emblem on a deep navy field

CreditSense AI

2025

A leak-free ML pipeline that scores loan eligibility — benchmarking five classifiers with stratified cross-validation to find the most reliable approve / reject model.

Pythonscikit-learnpandasNumPy
ml source

$ tail -n 3 blog.log

all posts

Aug 17, 2026

Human brain cells to run AI systems?

Companies have started using real human brain cells to run AI systems. San Francisco-based biotech startup Parasma has officially launched its new biological…

#ai-news#funding#ai