UDEMY / COURSE COMPLETION
SANHA TAHIR / DATA ENGINEER
Better decisions
start with data
you can trust.
Behind every confident decision is data that has earned that confidence. That’s where I come in. I take messy, fragmented source data through carefully designed pipelines and turn it into polished, gold-layer data marts—ready for people to explore, understand, and act on.
Sanha's portfolio assistantA friendly guide to her work
AI-generated answers can make mistakes. Chats expire after 30 minutes of inactivity. New chat deletes chat history. Questions and AI answers may be retained for up to 90 days to improve this assistant. Please avoid sharing sensitive information.
From complexity to clarity.SCROLL TO EXPLORE ↓
A FEW REASONS TO SAY HELLO
Does this sound familiar?
Drowning in data, searching for meaning?
If your company has mountains of data and no clear way to make sense of it, I can help turn the mess into models, metrics, and marts that answer real questions.
Pipelines running late—or losing your trust?
Consider me your pipeline doctor. I trace the symptoms, diagnose the root cause, and help get your data flowing reliably again.
Inmon or Kimball? Let's get philosophical.
If you want to debate modeling techniques, question a grain, or weigh the trade-offs of a warehouse design, I'm your philosophizing partner. Bring your strongest opinions—and your use case.
01 / SELECTED WORK
Built to be useful.
Real problems. Considered decisions.
Data systems that do their job.
Dimensional modeling / Reverse ETL
Customer usage, with the whole picture.
A daily data product that connects product activity to the customer accounts behind it.
Read the case study ↗Infrastructure / Data platform
A place to test. Confidence to ship.
A complete GCP test environment, from infrastructure and CDC to warehouse builds and alerts.
Read the case study ↗Incremental modeling / Performance
Every five minutes. Every detail matters.
Rethinking event processing with measured arrival latency, explicit windows, and deterministic deduplication.
Read the case study ↗Distributed processing / NLP research
Climate denialism, through a data lens.
A collaborative Georgetown research project combining AWS, PySpark, and NLP to explore climate discussion on Reddit.
Read the case study ↗02 / THE JOURNEY
Experience with purpose.
From public policy to production platforms.
September 2025–Present
Data Engineer · Transfr
Transfr builds VR career-training products for schools and workforce organizations. I develop the data systems supporting internal analytics, customer-facing reporting, and operational feeds to business tools.
- Build and maintain ingestion, dbt transformations, and Airflow workflows on BigQuery and Cloud Composer, including daily and five-minute processing paths.
- Model customer, product, and revenue data using explicit grains, hierarchy bridges, versioned models, contracts, and tests.
- Deliver customer-usage data to Salesforce and ClientSuccess, working with business teams to define metrics and validate their interpretation.
- Provision infrastructure and monitoring with Terraform, including a new test environment, CI/CD, access controls, and pipeline-failure alerts.
- Investigate silent data failures, reconcile missing records, and harden ingestion with retries, batching, validation, and backfills.
- Contribute to the FastAPI reporting service, code reviews, documentation, and engineer onboarding.
May 2024–September 2025
Data Engineer · World Bank Group, Data360
Built more than 15 Python and Kedro ETL pipelines for Data360, integrating diverse source datasets into standardized, reusable outputs.
- Transformed source data into SDMX-compliant formats to support consistent integration and downstream use.
- Worked with source creators to curate metadata and document pipelines, improving discoverability and interpretation.
- Implemented validation and anomaly checks that reduced data inconsistencies by 15%.
- Designed and iteratively tested five stakeholder-facing visualizations, increasing engagement by 16%.
- Contributed to Git-based review and reproducible development workflows.
January 2023–August 2024
Research Assistant · Georgetown University
Developed a GCP document-processing pipeline for more than 140,000 PDFs, converting documents into images for digitization and extraction.
- Combined image processing, a custom CNN, and Google Cloud Vision to extract information from handwritten forms.
- Used fuzzy matching and tiered joins in pandas to connect extracted names and identifiers with external datasets.
August 2020–January 2022
Data Scientist · Afiniti
Developed SQL and Python data workflows supporting machine learning and customer analytics.
- Automated ETL processes with Talend and Python, reducing pipeline completion time by 30%.
- Integrated records from multiple sources using tiered matching strategies, achieving an 80%+ matching success rate.
- Maintained historical CRM data with MySQL and SCD Type II tracking.
- Investigated data and model issues, and communicated findings through dashboards and analyses for internal teams and clients.
THE TOOLS BEHIND THE WORK
A considered toolkit.
Ingestion and integration
Python, REST APIs, PostgreSQL CDC, Estuary, Kedro, Talend
Warehousing and modeling
SQL, BigQuery, dbt, dimensional modeling, incremental models, snapshots, SCD Type II
Orchestration and infrastructure
Airflow, Cloud Composer, Terraform, Cloud Build, Cloud Storage, IAM, Secret Manager
Quality and delivery
Data contracts, freshness checks, reconciliation, SQLFluff, pytest, Git, FastAPI, reverse ETL
Distributed processing and AWS
Apache Spark / PySpark, AWS S3, SageMaker, Hadoop S3A, Parquet, DataFrames, UDFs
NLP and machine learning research
Spark NLP, Hugging Face Transformers, PyTorch inference, pretrained RoBERTa, Spark ML pipelines, logistic regression, gradient-boosted trees
Analysis and visualization
pandas, NumPy, Matplotlib, Seaborn, R, Shiny, Quarto, GitHub Pages
03 / CONTINUING TO LEARN
Certifications & learning.
Explore completed courses by topic.
A little curiosity goes a long way.04 / BEYOND THE PIPELINES
Engineer by trade.
Curious by nature.
My background spans customer analytics, public policy, and production data platforms. Across those settings, I enjoy the same challenge: understanding what people need from their data and building the systems that make it dependable.
I ask questions early—about definitions, ownership, edge cases, and the decisions a dataset should support. I like delivering something useful, learning from feedback, and improving it as the requirements become clearer.
I hold a master's in Data Science for Public Policy from the McCourt School of Public Policy at Georgetown University and a bachelor's in Economics and Mathematics from Lahore University of Management Sciences.
Away from the keyboard, music and dance are part of my routine. Zumba gets my day moving, and a burst of cardio helps me reset after a challenging one.