The Resume Foundry
How It Works

How The Resume Foundry works.

Use this page when you want the detailed pass: what each path does, what each export is for, when preview refreshes, and which settings are worth touching.

One Markdown source PDF / DOCX / TXT / QA bundle Preview before export
Sample output

The sample is a real render.

The sample uses a real founder resume, rendered with the same engine used for preview and export.

Sabbir Hossain
Atlanta Metropolitan Area, GA | Open to Relocation (US/Canada) | TN Visa - No Sponsorship Required
SUMMARY
Data Engineer with production ownership of mission-critical ETL and ELT pipelines serving enterprise analytics across 4+ operational systems. Delivered 83% query performance gains on 23M+ row datasets, 130K+ historical record recovery, and observability across 12+ pipelines. Strengths include Kimball data modeling, Airflow orchestration, backend API integration, and support for cloud modernization toward GCP and BigQuery.
SKILLS
Data Engineering: Python, SQL, Apache Airflow (DAGs), Apache Spark, PySpark, Teradata, PostgreSQL, ETL/ELT, data pipelines, Kimball data modeling, schema design, SCD Type 2, data quality, observability, monitoring, data lineage, data warehousing, data integration, batch processing, orchestration
Languages And Frameworks: Python, SQL, R, C, Bash, Java, JavaScript, TypeScript, Node.js, React, Flask, REST APIs, GraphQL APIs
Cloud, Warehousing, And Streaming: Apache Kafka, Snowflake, BigQuery, Amazon Redshift, dbt, AWS (S3, EC2, RDS), GCP (BigQuery, Cloud Composer), Azure
Engineering: Docker, Kubernetes (exposure), Linux, Git, CI/CD, GitHub Actions, Terraform (concepts), microservices, Agile/Scrum, technical documentation, JIRA, Confluence, TDD
Ml And Analytics: PyTorch, TensorFlow, scikit-learn, query optimization, anomaly detection, data structures and algorithms
EXPERIENCE
Bell Canada | Toronto, ON - Remote | Full Time
Data Engineer | Jun 2025 - Present
Owned the NTS data pipeline, a resilient 3-tier ETL/ELT architecture across staging, warehouse, and analysis layers on Teradata. Integrated SmartPath REST API, Maximo ERP, IPACT billing, and LDAP directory services using Python and SAS Data Integration while enforcing data contracts and Kimball dimensional modeling across DEV, QA, and PROD.
Delivered a 78-attribute MicroStrategy analytics cube with derived metrics and cross-filter interactivity. Secured director sign-off and production deployment within 6 weeks, and reduced build risk by using a SQL view instead of a physical fact table for event-to-request duration joins.
Reduced query latency 83% from 12 minutes to 2 minutes on a 23M+ row join by replacing runtime computation with a materialized pre-aggregation layer, eliminating production timeouts and stabilizing nightly SLA compliance.
Diagnosed systemic data drift caused by a misconfigured temporal filter, executed staged historical recovery totaling 130K+ records, and fixed 22K+ employee misattributions using SCD Type 2 temporal joins. Expanded analytical coverage from 1 month to 9+ months.
Built a Python and Airflow observability framework with configuration-driven checks for schema validation, anomaly detection, and threshold alerts across 12+ data pipelines. Reduced debugging time 60% and prevented 15+ monthly data quality incidents.
Promoted to Technical Gatekeeper within 3 months. Governed defensive coding standards and peer reviews, authored validation documentation including ERDs, data flow diagrams, and count-by-stage proofs, and supported modernization planning for a GCP and BigQuery migration targeting a snowflake schema.
Built a stateful sessionization algorithm in Python using two-pass group-by propagation to resolve event sequencing defects, achieving deterministic mapping across distributed agent sessions with zero QA calculation errors.
Served as designated backup owner for the SmartPath production API during a company-wide code embargo, managing escalation readiness and resolving critical data pipeline failures for 12+ dependent systems under time pressure.
Johns Hopkins University | Baltimore, MD - Remote | Part Time
Bioinformatics Software Development Research Assistant | Sep 2022 - Present
Engineered ETL pipelines processing 750+ TB of multi-omics data on HPC clusters using Python, R, and SQL, accelerating biomarker discovery 40% and reducing analysis time 40%.
Architected and maintained an open-source full-stack platform using Python, R, JavaScript, and C with microservices patterns and Docker. Reduced analysis load times 83% and supported 100+ global researchers.
Selected as 1 of 12 Harvard NCRC plenary speakers from 5,000+ applicants and authored 35-page research manuscripts with reproducible analyses. Also received Best Oral Presentation at ABRCMS 2023 among 6,500+ attendees.
University of Toronto | Toronto, ON - Hybrid | Part Time
Software Development Research Assistant | Sep 2019 - Apr 2024
Reduced analysis effort by 30+ hours per week across 7 research teams by building full-stack platforms in Python, R, C, and Java. Cut environment setup time 50% with Docker and improved UI render times 45% in Next.js.
Led Agile/Scrum adoption and mentored 5 junior developers, owning the full SDLC from requirements through production deployment and maintenance.
PROJECTS
NYC Taxi Data Engineering Pipeline | Repo: https://github.com/itsSabbir/nyc-tax-de
Built an end-to-end batch ELT pipeline with Postgres, dbt, Airflow orchestration, and Docker. Implemented medallion architecture with bronze, silver, and gold layers, star-schema marts, and automated data quality tests.
Image Processing Pipeline Server | Repo: https://github.com/itsSabbir/ImageProcessingPipeline
Built a multi-threaded C server with POSIX threads and sockets that handled 100+ concurrent clients at under 100 ms latency, improving throughput 30% through TDD and CI/CD practices.
Anomaly Detection System | Repo: https://github.com/itsSabbir/anomaly-detection-system-2
Built a full-stack platform with Node.js, Express, React, and TypeScript that orchestrated Python ML models including YOLOv5 and LSTM, with AWS S3, RDS, and Docker deployment.
MicrobiomeExplorer R Package | Repo: https://github.com/itsSabbir/MicrobiomeExplorer
Built an open-source modular R package for metagenomic analysis integrating ETL pipelines, Bioconductor-based statistics, and interactive Shiny visualizations.
EDUCATION
University of Toronto | B.Sc. (Hons) Computer Science, Bioinformatics and Computational Biology
Toronto, ON | 2024
GPA: 3.96/4.0
Relevant coursework: Data Structures and Algorithms, Database Systems, Distributed Systems, Machine Learning, Operating Systems, Software Design, Cloud Computing
AWARDS AND PRESENTATIONS
Poster Presentation
ABRCMS | 2024
Competed among 150+ graduate-level presenters.
Plenary Speaker
Harvard NCRC | 2024
Selected as 1 of 12 speakers from 5,000+ applicants.
Undergraduate Oral Presentation
ABRCMS | 2023
Ranked as a top presenter among 80 oral presenters at an event with 6,500+ attendees.
Friends of Arts and Science Awards
University of Toronto | 2022, 2023, 2024
Recognized across Computer Sciences and Physical and Life Sciences.
01 Main path

Create is the default path. Use it when you control the source or want to draft locally in the app.

02 Fallback path

Audit is for existing files. Upload a PDF or DOCX when the artifact already exists and you want parser-risk feedback first.

03 What stays fixed

The studio keeps one calm interface. Resume themes are optional and only affect the document accents, not the core workflow.

Outputs

What the bundle contains.

All four outputs come from the same source and the same render settings, so you are not managing separate versions.

PDF

The send-ready attachment. Use it when the application flow expects a final visual document.

DOCX

The primary portal upload when editable formats are allowed. This is usually the safest submission artifact for ATS parsing.

TXT

The plain-text extraction check. Use it to spot broken bullets, missing links, or ugly copy-paste behavior fast.

QA

The machine-readability report. It flags parser-risk, visible-link issues, structure problems, and general submission readiness.

Create Path

The normal workflow from source to bundle.

This is the path to use when you have Markdown already or want to draft from scratch in the browser.

01
Choose your source.

Use Upload Markdown if you already have a source file. Use Write in App if you want a local browser draft.

02
Load a starter only if it helps.

Starter templates are drafting accelerators. Clicking one loads starter content and recommended defaults into the local draft. It does not lock the renderer.

03
Generate Preview.

Preview is the inspection step. It renders the current draft with the current page and layout settings so you can review the actual page before export.

04
Open Advanced only when needed.

Tune Layout covers reset-to-template, resume themes, page size, link display, and the manual font-size and spacing controls. Most users should stay in the quick path until preview shows a real problem.

05
Render the bundle.

When the page looks right, export once. The bundle includes PDF, DOCX, TXT, QA, and the original Markdown source.

Preview refresh logic

If the draft or layout settings change after the last preview, the app marks the page as needing refresh. That is intentional. It prevents you from trusting an old page snapshot.

Audit Path

Use this when the file already exists.

Audit is for existing PDF or DOCX resumes. It is the fast answer to: “Is this artifact safe enough to ship as-is?”

01
Upload the real file.

The audit checks the same artifact a recruiter or portal would see, not a different source file that never gets submitted.

02
Review extraction and structure.

The audit looks at text recovery, headings, contact details, visible links, and broader structural risks that can make resumes hard to parse.

03
Decide the next move.

If the file is weak, rebuild it in Create. If it is strong, keep the artifact and apply only the fixes that matter.

Details

A few practical details.

The sample viewer

The sample is a real founder resume rendered through the same live preview engine and fixed styling used for preview and export. It proves the output on a real document, not a mockup.

Why the chrome stays restrained

The interface stays restrained so the workflow remains readable. The resume itself can still use conservative color themes for section titles, rules, and links when you want a little more character.

What stays local

The Write in App draft stays in local browser storage on this device. It is there so you can iterate without juggling files while still exporting from one source of truth.

Why visible links stay on by default

Visible URLs are more reliable for ATS parsing, plain-text extraction, and recruiter copy-paste. Collapsed labels are available for the cleaner visual pass, but they are opt-in.

FAQ

Should I start in Create or Audit?

Start in Create if you have Markdown or want to draft locally. Use Audit only when the resume already exists as a PDF or DOCX and you want to evaluate that artifact first.

What does “Preview needs refresh” mean?

It means the draft or settings changed after the last preview. Generate preview again so the page you inspect matches the current state.

Do the preview and export match?

Yes. Preview uses the same fixed styling, page size, and layout settings that the export step uses.

Why keep both PDF and DOCX?

PDF is the final visual artifact. DOCX is often the safer upload for portals and ATS parsing. Keeping both gives you the clean visual version and the recruiter-system-friendly version.