Analyse
SQL against high-volume transaction data, then finding the pattern that actually explains the alert rather than the one that looks tidy.
Open to data science & analytics roles
Hello, I'm
K.M. Naimuddin
A Data analyst, Engineer and Applied ML researcher.
- Data Analyst
- Data Science
- Engineer
- Researcher
- Problem Solver
I spend my days on financial-crime analytics at SEBPO: SQL against high-volume transaction data, and the Power BI reporting that compliance teams run their week from. The rest of my time goes to deep learning — road accident classification from CCTV, now under review at ICCIT 2026, and a study on how fraud models rot as behaviour shifts.
- ~2 yrsAML & KYC analytics
- 10,000+cases reviewed
- 96.9%research accuracy
- 2research papers
Dhaka, BD
About
A bit of background
Build
Power BI reports and automated pipelines that a compliance team opens every morning, not dashboards nobody asked for.
Research
Deep learning on problems with messy real-world data: accident detection from CCTV, and model decay in fraud systems.
I work as a data analyst at ServiceEngine Limited (SEBPO) in Dhaka, on a team doing compliance analytics for financial-sector clients. Day to day that means SQL against high-volume transaction data, cleaning and reconciling customer records that rarely match cleanly, and building the Power BI dashboards that compliance officers and team leads actually run their week from.
When a screening decision has to survive an audit, you learn to document why the data says what it says, not just what it says.
That habit carried into everything else I build. Before SEBPO I finished a Computer Science degree at Premier University, Chattogram, where my thesis became a deep-learning system for classifying road accidents from CCTV footage. That paper is under review at ICCIT 2026, and I'm now looking at what happens to fraud-detection models when the behaviour underneath them changes.
- Current role Data Analyst, SEBPO
- Location Dhaka, Bangladesh
- Experience ~2 years
- Education B.Sc. CSE
- Domain Finance, AML, KYC, Transaction Monitoring
- Research Interests Data Science, Deep Learning, Agentic AI, Computer Vision
- Nationality Bangladeshi
Languages
- Bengali Native Mother tongue
- English B2 IELTS Academic · WSDA Professional Communication, grade A+
- German A1 Self-study, in progress
- Hindi B2 Working Proficiency
Toolkit
Skills
Grouped by what I reach for, rather than scored out of a hundred.
Programming & query
Data analysis
BI & visualisation
Databases & pipelines
Machine learning
Tools & platforms
Domain knowledge
Experience
Where I've worked
Jan 2025 – Present
Data Analyst
ServiceEngine Limited (SEBPO) · Management Analytics & Research Solutions · Dhaka
- Analyse high-volume customer and transaction datasets for financial-sector clients, writing SQL and preparing around 300 records a day for compliance review.
- Run trend and exception analysis across transaction monitoring alerts, identifying suspicious activity patterns and the false-positive drivers that inform screening thresholds. Repeat escalations fell 18%.
- Build and maintain the Power BI dashboards and recurring KPI reports covering case volume, turnaround time and quality metrics, used daily by client compliance teams and internal management.
- Reconcile customer identity records across partially matched source databases, resolving data-quality conflicts and documenting defensible determinations for audit and regulatory review.
- Present findings and escalation summaries to client compliance officers and team leads, translating technical detail into recommendations they can act on.
- Execute KYC, CIP and sanctions screening workflows including PEP and adverse media checks, holding a 98% quality-assurance accuracy rating across 10,000+ cases.
Oct 2024 – Dec 2024
Data Analyst, Intern
ServiceEngine Limited (SEBPO) · Dhaka
- Extracted, cleaned and validated customer datasets supporting KYC screening, processing roughly 1,200 records a week.
- Produced the daily and weekly reports on screening throughput and data-quality exceptions for team leads.
- Retained as a full-time data analyst at the end of the internship.
Jun 2024
Data Analyst, Intern
Mentorness · Remote
- A one-month analytics internship on the proof-of-work platform, working through assigned datasets end to end.
- One line on what you actually built here would help — a dataset, a dashboard, a model.
Oct 2022 – Dec 2023
Technical Head
HophyCare · Health-tech startup · Bangladesh
- Ran the technical side of an early-stage startup building a home healthcare platform, alongside the final years of my degree.
- What did you own — the web app, the database, a small team? One or two specifics would carry this entry.
Feb 2022 – Jul 2022
Graphic Designer, Intern
Youth School for Social Entrepreneurs (YSSE) · Youth-led non-profit
- Six months producing campaign and event graphics for a youth-led social enterprise organisation.
Education
-
B.Sc. in Computer Science & Engineering
Premier University, Chattogram · CGPA 3.39 / 4.00 Feb 2019 – Feb 2024
Enrolled spring 2019; final results published 20 February 2024, the duration extended by the nationwide COVID-19 university closures. Core coursework in machine learning, computer vision, statistics, data structures and algorithms, database systems and software engineering.
Leadership & activities
Feb 2023 – Feb 2024
Vice-Chair
IEEE Premier University Student Branch
Second in command of the university's IEEE branch, running the technical event calendar and the volunteer team behind it.
Sep 2022 – Dec 2023
General Secretary
CSE Club, Premier University
Ran the department's student club: membership, meeting cadence and the workshops and contests it put on across the year.
Jan 2022 – Dec 2023
Technical Head
Robotics Club, Premier University
Led the technical side of the robotics club, mentoring junior members through build projects and competition entries.
2022 & 2023
Organiser & Volunteer
Bangladesh Olympiad in Informatics
Helped run two national programming olympiad rounds hosted at the Premier University CSE department.
2019
Organiser
Premier University IT Fest
On the organising team for the university's annual technology festival.
Jan 2023 – Present
Member
IEEE · no. 99076130
Also a member of the Leo Club of Chittagong Karnaphuli since December 2022, on community service projects.
Research
Research & publications
Applied machine learning on problems where the data misbehaves: video that has to be labelled by hand, and models that quietly stop working.
Under review · ICCIT 2026
Computer Vision Based Road Accident Classification from Traffic Surveillance
A hybrid deep-learning model that decides whether a stretch of CCTV footage contains a road accident. MobileNetV2 pulls visual features out of each frame, a bidirectional LSTM reads the frames as a sequence, and the pair reach 96.9% accuracy on a dataset we collected ourselves: 126 real surveillance videos, about 1,890 sampled frames. Building the dataset was most of the work, including diagnosing and correcting a daylight sampling bias that was suppressing low-light performance.
Paper link after acceptance
In progress
Detecting Concept Drift in Machine-Learning-Based Transaction Fraud Detection
Fraud models are trained once and then quietly decay, because the behaviour they were fitted to keeps moving. This study measures how far a static model falls behind one that is periodically retrained, and whether drift-detection methods raise the alarm before predictive performance has already degraded in a way anyone would notice.
Write-up coming
Work
Projects
Analytics, machine learning and a bit of web engineering. Code is on GitHub and Kaggle.
Kaggle notebook
Autonomous Traffic Sign Recognition
A convolutional neural network for multiclass traffic sign classification, reaching 95% accuracy through image preprocessing, data augmentation and hyperparameter tuning. Evaluated with confusion matrices and per-class metrics.
Open source
Bank Loan Insights Analysis
SQL-based analytical pipelines feeding an interactive Power BI dashboard for loan portfolio risk: approval rates, default exposure and portfolio performance KPIs. The recurring reporting is automated, replacing manual spreadsheet consolidation.
Open source
Paris 2024 Olympic Insights
Athlete, event and medal tables for the Paris 2024 Games cleaned in Python and modelled into a single Power BI report, with DAX measures for medal counts by country, sport and discipline.
Open source
PUAIS — Enrollment Module
The course enrollment module for the Premier University Academic Information System, with automated validation of credit-limit and section-capacity constraints, real-time notifications and faculty override support.
Open source
Pizza Sales Analysis
Order-level sales data queried in SQL Server, then a Power BI dashboard covering revenue, order volume, average order value, busiest hours and days, and best and worst sellers by category and size.
Open source
HR Analytics Dashboard
Employee data reshaped in Power Query and turned into an attrition dashboard: headcount, leaver rate and the breakdown by department, age band, salary band and years of service.
Nothing in that area yet.
There's more on GitHub
Smaller experiments, practice datasets and the raw report files all live there.
Browse the repositoriesRoad Accident Classification from Traffic Surveillance
Each video is sampled into a fixed-length frame sequence. A time-distributed MobileNetV2 extracts visual features per frame, those are flattened and fed to a bidirectional LSTM that learns the temporal pattern, and a dense layer with dropout makes the accident / non-accident call. Transfer learning does the heavy lifting on a dataset this small.
Bank Loan Insights Analysis
Three pages: a summary with the headline lending KPIs, an overview with the trend and regional cuts, and a details table. Every KPI was written first as a SQL query and then reproduced as a DAX measure, which made the numbers easy to check against each other.
Paris 2024 Olympic Insights
The raw files came as separate athlete, event and medal tables. Python handled the cleaning and the joins; the model and every measure were built in Power BI with DAX. The report pages break medals down by country, sport and discipline.
Pizza Sales Analysis
The questions came first as SQL against SQL Server: total revenue, average order value, pizzas per order, and the busiest hours and weekdays. The Power BI pages then put those alongside best and worst sellers by revenue, quantity and order count, split by category and size.
HR Analytics Dashboard
Most of the work was in Power Query: the source spreadsheet had inconsistent categories and a few unusable columns. Once it was clean the report shows headcount and attrition rate, then cuts them by department, age band, salary band and years at the company.
Credentials
Certifications
Every entry links to its verification page.
Data Analytics Bootcamp 3.0
Codebasics
Certificate link needed
PHP with Laravel
BASIS Institute of Technology & Management
Public link needed
Professional English Communication
WSDA · Grade A+
Public link neededReferences
People who can vouch for me
Contact details are on request rather than in public.
Prof. Dr. Shahid Md. Asif Iqbal
Professor & Associate Dean, Faculty of Engineering
Premier University
Mohammad Hasan
Assistant Professor, Dept. of CSE
Premier University · Moderator, Programming Wing, PU Computer Club
Tusar Karmoker
Assistant Program Manager
ServiceEngine Limited (SEBPO)
Contact
Get in touch
Email is the fastest way to reach me.