Work

The whole timeline.

Three years at Gigalogy across SmartAds, maira-api and R&D, two publications, and the study that came before and after.

Oct 2026 – PresentCurrent

MSc in IT Engineering

FH Wedel · Hamburg, Germany

Postgraduate study in IT Engineering.

Feb 2026 – Sep 2026

Software Engineer (R&D)

Gigalogy / R&D · Dhaka, Bangladesh

Rebuilt Maira’s retrieval architecture from vector RAG to graph RAG, over a purpose-built knowledge graph with a live memory layer.

8 details
  • Evaluated database engines for a growing customer base — benchmarking Tsurugi DB against alternatives on structure, implementation cost and scaling behaviour.
  • Rebuilt Maira AI from the ground up: vector RAG replaced by graph RAG over a purpose-built knowledge graph.
  • Designed the knowledge graph for multi-tenancy, access scoping, and dynamic creation and deletion at the lowest cost per tenant.
  • Built a memory layer that ingests conversations live, so context accumulates instead of being re-derived per session.
  • Compared local and hosted models, and RST parsers, to reach a state-of-the-art extraction pipeline.
  • Designed a query router that splits traffic between vector RAG and graph RAG to cut cost and latency without losing answer quality.
  • Enabled graph RAG to reference multiple dataset graphs simultaneously within a single query.
  • Researched discourse conflict resolvers that engage automatically to reconcile contradictory narratives across sources.
Knowledge GraphsGraph RAGVector DatabasesTsurugi DBRST ParsingPythonMulti-tenancy
Oct 2025 – Feb 2026

Software Engineer

Gigalogy / maira-api · Dhaka, Bangladesh

Built realtime voice AI over customer datasets, and shipped research and deep-thinking capabilities into the Maira assistant.

5 details
  • Integrated research and deep-thinking features into the Maira assistant.
  • Built realtime live voice chat with AI connected to customer datasets — users ask questions about their own company and get accurate spoken answers live.
  • Held the voice pipeline to millisecond response times: turn detection, voice activity detection (VAD), and noise cancellation.
  • Implemented intent detection and query routing so each request took the cheapest path that could still answer it.
  • Reused retrieved information and conversation context across turns to cut redundant model calls and hold cost down.
Realtime Voice AIVADTurn DetectionNoise CancellationIntent DetectionQuery RoutingPython
Mar 2024 – Oct 2025

Software Engineer

Gigalogy / SmartAds · Dhaka, Bangladesh

Performance and throughput work on the ad delivery pipeline — measured before and after, and written up as a case study.

3 details
  • Improved API performance by up to 99.7% on the worst-performing paths.
  • Reduced the API calls required for a single operation from 10,000 to 140.
  • Cut a batch job’s runtime from 2 hours to 3 minutes.
PythonAPI OptimizationBatch Processing
2023Preprint

Deep Fake Analysis Using Convolutional Neural Network and Vision Transformer

Preprint · With Mohammad Monirujjaman Khan

Dual-stage deepfake detection: a fine-tuned EfficientNet extracts frame-level features, and a Vision Transformer classifies the resulting sequence for spatial and temporal manipulation artifacts.

Computer VisionVision TransformersEfficientNetDeepfake Forensics
Nov 2022Published · IEEE

Liver Cirrhosis Prediction using Machine Learning Approaches

IEEE UEMCON 2022 · With Mohammad Monirujjaman Khan

Comparative diagnostic study training SVM, Random Forest and Logistic Regression classifiers on clinical metrics to predict cirrhosis staging, with feature selection tuned for early-stage detection.

Machine LearningMedical DiagnosticsFeature Engineering
2017 – 2021Education

BSc in Computer Science and Engineering

North South University · Dhaka, Bangladesh

Concentration in Intelligent Systems and Machine Learning. Graduated with honors.

1 details
  • Senior design project: an autonomous self-driving vehicle prototype combining convolutional lane detection and barrier classification with ultrasonic and IMU sensor fusion on Raspberry Pi and Arduino.
Sensor FusionComputer VisionMicrocontrollers
2014 – 2015Education

Undergraduate studies in Computer Engineering

North Dakota State University · Fargo, North Dakota, USA

Foundational coursework in C++, electrical circuits, microcontroller systems and physics.