Software Engineer @ Adfinix · Dhaka, Bangladesh

Mehraz Hossain
Rumman

I build the backends behind real-time ad-tech: bidding, click and impression pipelines, and analytics on Django and ClickHouse. Google Summer of Code 2025 contributor with IOOS. Patches merged in Django, DRF, Scrapy, Celery and Kombu. On the side I analyse Bangladesh's public data, from road accidents to dengue.

Developer ID
Portrait of Mehraz Hossain Rumman

Mehraz Hossain Rumman

Software Engineer · Adfinix

Dhaka, Bangladesh · --:--

Focus
Backend · Ad-tech · Data
Experience
4+ years
Open source
7 projects incl. Django, DRF, Celery
Highlight
GSoC 2025 · Keynote, Hacking Limnology
  • 59 repos
  • 18 followers
github.com/MehrazRumman

Now

Software Engineer at Adfinix

Building Adsloft and Aifinix: DSP/SSP tracking, OpenRTB bidding flows, and real-time analytics for advertisers and publishers. Since July 2024.

  • Django
  • DRF
  • ClickHouse
  • MongoDB
  • OpenRTB
  • Next.js
4+years shipping
2backends from scratch
7OSS projects patched
’25GSoC · IOOS

About

Mathematics graduate turned backend engineer. I spent my degree in programming contests, which is where I learned to care about correctness under pressure. Now I build ad-tech systems by day and contribute to the Python ecosystem I depend on. Through GSoC I shipped HTTP caching to pyOBIS, then took the work on stage as a keynote speaker at Hacking Limnology 2025. I've taught Python online since 2022.

Education

B.Sc. Mathematics

Jagannath University, Dhaka
2019 – 2024 · GPA 3.59 / 4.00

Coordinates

23.8103° N
90.4125° E

UTC+6 · open to remote

  • Python
  • Django
  • Django REST Framework
  • ClickHouse
  • MongoDB
  • PostgreSQL
  • Go
  • Next.js
  • React
  • FastAPI
  • Docker
  • Jenkins
  • GitHub Actions
  • OpenRTB
  • scikit-learn
  • Pandas
  • xarray

01 — Experience

Where I've worked

  1. Jul 2024 — PresentDhaka, Bangladesh

    Software Engineer Adfinix

    • Built the Adsloft backend from scratch: a SaaS that tracks DSP/SSP clicks, impressions and views and generates real-time data and detailed reports for advertisers and publishers.
    • Built the Aifinix backend from scratch with Django and DRF, designing scalable APIs, authentication, analytics pipelines and business logic; contributed to its frontend too.
    • Designed and delivered an analytical dashboard end to end: schema design, backend aggregation and frontend visualization for real-time insights.
    • Upgraded the Adfinix product from Django 3.2 to 5.1 with its dependencies, fixing bugs and optimizing hot paths along the way.
    • Implemented CPM and frequency capping, built rich media ads and resolved API issues across the platform.
    • Worked with OpenRTB for real-time bidding and built a mock server to test DSP and SSP request-response interactions.
    • Used MongoDB for high-volume event and analytics data, wrote a Slack bot for internal automation, and built the Adsloft marketing site in Next.js.
    • Django
    • DRF
    • ClickHouse
    • MongoDB
    • Next.js
    • OpenRTB
    • Docker
    • Jenkins
  2. May 2025 — Jul 2025Remote

    Google Summer of Code Contributor IOOS

    • Implemented HTTP-level caching in the pyOBIS Python package using requests-cache, with cross-platform cache paths, unit tests and docs. Every PR merged upstream.
    • Built a seagrass habitat analysis notebook combining OBIS and GBIF occurrences with environmental variables: One-Class SVM anomaly detection, temporal analysis and Species Distribution Model mapping.
    • Keynote speaker and workshop instructor at Hacking Limnology 2025 on pyOBIS and marine ecological modeling.
    • Python
    • requests-cache
    • Pandas
    • scikit-learn
    • xarray
    • Jupyter
  3. Nov 2022 — PresentRemote · Freelance

    Online Python Trainer

    • Teach Python from beginner to intermediate level, with a focus on debugging, reading errors and problem-solving.
    • Introduce core data structures and algorithms; run weekly exams and coding challenges to track progress.
    • Python
    • DSA
    • Mentoring
  4. Apr 2022 — May 2024Dhaka · Remote

    Trainee Software Developer TrendBreaker

    • Built a desktop application with PyQt5, and a React Native mobile app backed by Django.
    • Worked on data analysis projects to support business decisions.
    • PyQt5
    • React Native
    • Django
    • Data analysis

02 — Projects

Things I've built

Work · Adfinix

Adsloft

SaaS tracking DSP/SSP clicks, impressions and views with real-time reporting. Backend built from scratch, ClickHouse for event analytics, marketing site in Next.js.

  • Django
  • ClickHouse
  • Next.js
Work · Adfinix

Aifinix

Ad-tech platform backend: scalable APIs, auth, analytics pipelines and an end-to-end analytical dashboard over high-volume MongoDB event data.

  • Django
  • DRF
  • MongoDB
Research · GSoC

Seagrass Habitat Analysis

Merges OBIS and GBIF occurrences with environmental layers: One-Class SVM anomaly detection, temporal trends and SDM maps. Taught at Hacking Limnology 2025.

  • scikit-learn
  • xarray
  • SDM
Full-stack

KachaMorich

E-commerce app with accounts, payments, profiles and order reviews for customers, plus user, product, role and order-status management for admins.

  • Django
  • React
  • PostgreSQL
Academic

DSP with Fast Fourier Transform

Digital signal processing study applying the FFT to analyse and filter signals, built during my mathematics degree.

  • Python
  • NumPy
  • Signal processing
Notebook · Colab

pyOBIS Workshop Notebook

Hands-on Species Distribution Modeling with OBIS data, as run for the Hacking Limnology 2025 workshop. Open it in Colab and run it end to end.

  • Python
  • pyOBIS
  • Jupyter

03 — Data analysis

Public data, cleaned and mapped

Open, reproducible studies of Bangladesh built from government reports and press data. Every row keeps a link to its source.

Interactive Bangla map of road deaths by division in Bangladesh, with accident-prone highway spots marked Live map
Jupyter · Road safety2021 – 2026

Bangladesh Road Accidents

Road Safety Foundation and official BRTA monthly reports, transcribed from scanned PDFs, checked against each report's totals and analysed by year, vehicle, division and season. Ships with an interactive Bangla map of all 64 districts and 215 accident-prone highway spots.

7,359
road deaths in 2025
36%
of them on motorcycles
+26–51%
deaths per day during Eid travel
Chart of road deaths per year, 2021 to 2026 Heatmap of road deaths per day by month and year
  • pandas
  • matplotlib
  • seaborn
  • Leaflet
  • Jupyter
Maps of dengue cases per 100,000 people by division in Bangladesh, 2025 compared with 2026 Notebook
Jupyter · Public health2000 – 2026

Dengue in Bangladesh

Where dengue strikes, how the hot zones shift and who dies. Joins DGHS surveillance, district case counts, census population and climate data into hotspot maps, case fatality rates, seasonality and temperature-lag correlations.

116,990
cases in 2026, second-worst year
3.8×
Barishal vs national rate, 2025
~2×
fatality rate for women vs men
Heatmaps of dengue incidence and relative risk by division, 2020 to 2026 Weekly dengue cases in 2026 compared with 2025
  • pandas
  • GeoPandas
  • NumPy
  • matplotlib
  • Jupyter

04 — App development

On the Play Store, and on the way

One app live on Google Play and two more in active development.

Live on Google Play
Strat Roulette app icon

Strat Roulette

Spin the wheel and get an alternative strategy to shake up your next match. Supports CS2, Rainbow Six Siege, Fortnite, Overwatch and PUBG.

  • Android
  • Gaming
  • Published
In development

Becho

A business toolkit for merchants and small business owners: track sales, inventory and customers from a single app.

  • Android
  • Merchants
  • SMB
In development

Nirbhor

A medicine alarm app that reminds you and the people you care for to take the right dose at the right time.

  • Android
  • Health
  • Reminders

05 — Web scraping

Turning websites into datasets

Crawlers and extraction pipelines that ship clean, structured data from messy sites.

Client · 24baba.com3,700+ listings
24baba logo

Car listings & images

Scraped 3,700+ car listings with full specs and every photo, normalised into a clean dataset ready to import into the marketplace.

  • Python
  • Image pipeline
  • Deduplication
Combat sports7 sources

Fight results database

Event, bout and fighter results across five disciplines, unified into one schema from sites with very different layouts.

Healthcaredataset

Medicines catalogue

Drug names, generics, manufacturers, dosage forms and prices collected from pharmacy and regulator sites into a searchable catalogue.

  • Python
  • Structured data
  • Pharma

06 — Open source

Patches in the tools I use daily

07 — Skills

What I reach for

Languages

  • Python
  • Go
  • SQL
  • C
  • C++
  • Fortran

Frameworks

  • Django
  • Django REST Framework
  • FastAPI
  • Next.js
  • React
  • React Native
  • PyQt5

Data stores

  • PostgreSQL
  • ClickHouse
  • MongoDB

Infrastructure

  • Docker
  • CI/CD
  • GitHub Actions
  • Jenkins
  • Git
  • Bitbucket
  • OpenRTB

ML & analysis

  • Pandas
  • NumPy
  • GeoPandas
  • scikit-learn
  • xarray
  • matplotlib
  • seaborn
  • folium
  • Leaflet
  • Jupyter
  • SVM / One-Class SVM
  • Random Forest
  • Regression
  • Species Distribution Modeling

Mathematics

  • Linear algebra
  • Calculus
  • Fuzzy mathematics
  • Data structures & algorithms

08 — Competitive programming

Contests & hackathons

Max-rated Specialist (1425) on Codeforces as Izuku__Midoriya (1,750+ problems solved) · 3★ on CodeChef (keai33, 138 problems) · problems on LeetCode and HackerRank.

09 — Talks & writing

Said out loud

10 — Teaching & mentoring

Teaching programming & AI

One-on-one sessions across time zones — from first lines of code to contest algorithms to putting AI to work in a product team.

Programming

2 students · United States

Taught programming from the ground up: language fundamentals, problem decomposition, and building small projects end to end.

Competitive programming

4 students · Bangladesh

Coached data structures, algorithms, and contest strategy with weekly problem sets and upsolving on Codeforces-style problems.

AI consultancy

Project manager · United States

Advised on where AI and LLMs fit in a product roadmap, how to scope pilots, and how to evaluate results before committing.

11 — Contact

Let's build something.

Backend roles, freelance projects, open-source collaboration, or a question about GSoC. My inbox is open.

Intro — watch on YouTube