Md Robiuddin
Data Engineer with 4+ years building production data pipelines, retrieval-augmented AI systems and large-scale geospatial data products. M.Sc. researcher in time-series forecasting at Trent University, Peterborough, Canada.
Experience
Graduate Researcher
Jan 2025 – presentTrent University, Peterborough, ON · M.Sc. thesis in multivariate time-series forecasting- Study when and how information from other channels improves a forecast, using matched with/without-context comparisons across 10 benchmark datasets.
- Designed Target-Queried Channel Attention (TQCA), a dual-stream Mamba model that scores and gates context channels per target; context gains of up to 13.9% (humidity) and 12.7% (PJM electricity prices). Published at IEEE ICTAI 2026.
- Developed a learned Granger-style gate for TimeXer and iTransformer that keeps only the useful sources per target, beating fixed graphs of the same size in 65 of 65 paired comparisons.
- Run large experiment campaigns in PyTorch on Digital Research Alliance of Canada GPU clusters with Ray Tune and SLURM.
Data Engineer
Nov 2025 – Oct 2026Hjemla AS, Norway- Built a real-time RAG pipeline (FastAPI) that turns property sales PDFs into chunked text and embeddings for semantic retrieval and structured AI outputs, with a database-backed cache that cut repeat-request latency.
- Engineered a PostGIS + dbt geospatial pipeline computing distances between 4M+ properties and risk and noise zones, reducing processing time from months to days.
Data Engineer
Aug 2021 – Oct 2025Softwrd Ltd, Dhaka (data engineering team of Placepoint AS, Norway)- Led a team of 3 delivering modular, Airflow-orchestrated ELT pipelines deployed with Docker and AWS Lambda.
- Built reusable data models in SQL, Python, dbt and PySpark for analytics, marketing and ML use cases; refactored legacy pipelines for a 40% performance gain.
- Introduced data validation and monitoring, CI/CD, code review and testing practices across the team.
Graduate Teaching Assistant
Jan 2025 – presentTrent University · Software Design & Modelling, Artificial IntelligencePython Programming Instructor
Feb 2024 – Dec 2024Center for Development of IT Professionals · mentored 100+ studentsSelected projects
LLM-powered document analyzer
Real-time RAG over unstructured sales documents: PDF parsing, chunking, embeddings, ranked retrieval and result caching.
Geospatial risk proximity
Closest-distance computation for 4M+ properties against risk and noise zones with optimized spatial joins and incremental models.
Real-estate & automotive ELT suite
Modular pipelines for daily listings, new-build trackers and company pages feeding dashboards and data products.
softwrdwrangler
Open-source Python library on PyPI: URI-based S3 read/write for CSV, JSON, Pickle and GeoJSON in cloud-native pipelines.
Research & publications
Target-Queried Channel Attention: Measuring and Exploiting Cross-Channel Context in Multivariate Time-Series ForecastingIEEE ICTAI 2026 · DOI 10.1109/ICTAI70596.2026.00255
Skills
PythonSQLApache AirflowdbtPySparkPostgreSQL / PostGISMongoDBAWS (S3, Lambda)DockerCI/CDFastAPIRAG & embeddingsvLLMPyTorchTime-series forecasting