{ "name": "Hiwa Feizi", "title": "Data Engineer", "location": "Dordrecht, Netherlands", "email": "hiwafeiziii@gmail.com", "links": { "linkedin": "https://www.linkedin.com/in/hiwa-feizi-a9a620167/", "github": "https://github.com/hiwafeizi" }, "summary": "Data engineer at Allianz Netherlands. Driving the technical direction of a SAS-to-Snowflake migration. Leading an external 10+ person delivery team, owning the core upstream layer of the Data Office, and setting the AI tooling patterns the wider data org now uses across multiple country offices.\n\nStrong in dbt, Snowflake, Python, SQL, and the surrounding cloud stack. The work I'm proudest of is the kind that makes whole teams faster. Automation that compounds, tooling other people adopt, decisions that hold up over time.\n\nBackground in cognitive science and AI. Before Allianz, shipped two production products solo on GCP with CI/CD (NeoCru, LinkInLead).", "traits": [ "Structured, creative, and multidisciplinary engineering mindset", "Fast learning with a focus on clean abstractions", "Clear cross-functional communication" ], "domains": [ { "id": "languages", "label": "Languages", "description": "Programming and query languages I work in.", "skills": [ "Python", "SQL", "JavaScript", "C++" ] }, { "id": "data_warehouse", "label": "Data & Warehouse", "description": "Data platforms, formats, and engineering practices.", "skills": [ "dbt", "dbt Cloud", "Snowflake", "Snowflake API", "SAS", "PostgreSQL", "Oracle SQL", "Power BI", "Mainframe PD binary", "Zero-copy clones", "ETL pipelines", "Batch processing", "Data modeling", "Large-scale migrations", "Data validation", "Parity testing", "Runtime checks", "Data quality", "Metadata automation", "Streaming", "Parquet", "CSV", "JSON", "XLSX", "YAML" ] }, { "id": "cloud_devops", "label": "Cloud & DevOps", "description": "Cloud providers and delivery tooling.", "skills": [ "GCP Cloud Run", "GCP Cloud Storage", "GCP Cloud Build", "GCP Cloud Functions", "GCP Cloud Tasks", "GCP reCAPTCHA", "GCP Logs", "Azure ADP", "Azure Blob Storage", "CI/CD (GitHub Actions)", "Containerization", "Git", "GitHub", "Jira" ] }, { "id": "software_engineering", "label": "Software Engineering", "description": "Frameworks, internal tooling, and engineering practices.", "skills": [ "Flask", "API design", "Internal tooling", "Automation", "Threading", "Binary parsing", "Documentation" ] }, { "id": "ai_ml", "label": "AI / ML", "description": "LLM tooling, ML libraries, and applied AI.", "skills": [ "OpenAI API", "Claude Code", "Transformers", "PyTorch", "scikit-learn", "pandas", "NumPy", "Dlib", "RobBERT", "T5", "LightGBM", "NLP", "Computer vision", "Feature engineering", "Neural networks", "Elastic Net regression", "LLM API chaining", "Prompt engineering", "AI-assisted development" ] }, { "id": "working_style", "label": "Working Style", "description": "How I work with teams and on engineering decisions.", "skills": [ "Code review", "Technical leadership", "Mentorship", "Cross-team coordination", "Analytical mindset", "Clear cross-functional communication" ] } ], "experience": [ { "company": "Allianz", "location": "Rotterdam, Netherlands", "timeframe": "Feb 2025 - Present", "roles": [ { "title": "Data Engineer", "timeframe": "Jul 2025 - Present", "bullets": [ { "text": "Acted as Allianz-side technical lead for a 10+ person IBM delivery team on the SAS-to-Snowflake migration: owned acceptance criteria, dbt code review, and model-design review; coordinated across IBM, SAS, IT, and the Data Office; mentored a junior engineer on the Allianz side.", "project": "SAS to Snowflake migration", "skills": [ "Technical leadership", "Code review", "Cross-team coordination", "Mentorship", "Clear cross-functional communication", "dbt", "Snowflake", "SAS", "Large-scale migrations" ] }, { "text": "Defined the SAS team's tasks throughout the migration and gave them targeted dbt/Snowflake training as needed, enabling them to validate parity from the SAS side.", "project": "SAS to Snowflake migration", "skills": [ "Cross-team coordination", "Mentorship", "Clear cross-functional communication", "dbt", "Snowflake", "SAS", "Data validation", "Parity testing" ] }, { "text": "Owned technical decisions on pstag, the Data Office's upstream source: reviewed pull requests, validated results against data, set the team's PR standards, and proposed enhancements to the project.", "project": "pstag ownership", "skills": [ "Technical leadership", "Code review", "dbt", "Snowflake", "Data validation", "Data quality" ] }, { "text": "Defined the SAS-migration source-loading strategy and co-designed new ingestion processes with IT for previously missing sources.", "project": "SAS to Snowflake migration", "skills": [ "SAS", "Snowflake", "Large-scale migrations", "Data modeling", "Cross-team coordination", "ETL pipelines" ] }, { "text": "Built a generic Python parser for the mainframe PD binary format after spotting data loss in the prior CSV ingestion. Switched input to binary while keeping CSV output, added batching and structured logging, and moved per-table config to JSON files in the repo so changes ship through version control instead of IT tickets.", "project": "PD binary parser", "skills": [ "Python", "Mainframe PD binary", "Binary parsing", "ETL pipelines", "Batch processing", "Automation", "Internal tooling", "JSON", "CSV", "Data quality" ] }, { "text": "Built an AI-driven SAS↔Snowflake column matching pipeline: chains LLM API calls with live SAS and Snowflake connections, parses metadata, samples row-level values, and validates matches programmatically. Reached 93% average accuracy and became what the team uses to validate the migration and review dbt developers' PRs.", "project": "AI column matching pipeline", "skills": [ "LLM API chaining", "Prompt engineering", "AI-assisted development", "OpenAI API", "Python", "SAS", "Snowflake", "Data validation", "Parity testing", "Automation", "Code review" ] }, { "text": "Crawled SAS server directories to extract transformation logic from 1,000+ SAS files, cataloguing reusable code to accelerate the migration of legacy pipelines to dbt on Snowflake.", "project": "SAS to Snowflake migration", "skills": [ "SAS", "dbt", "Snowflake", "Large-scale migrations", "Data modeling", "Python", "Automation" ] }, { "text": "Migrated pstag from script-based logic to dbt models and YAML configurations; programmatically generated ~660 YAML files and 2,000+ SQL models using Python and Snowflake metadata.", "project": "pstag migration", "skills": [ "dbt", "YAML", "SQL", "Python", "Snowflake", "Data modeling", "Large-scale migrations", "Metadata automation" ] }, { "text": "Built a validation framework using dbt macros and runtime checks, achieving 100 percent row- and column-level parity against production for daily data loads exceeding 50 million rows.", "project": "pstag migration", "skills": [ "dbt", "Data validation", "Parity testing", "Runtime checks", "Data quality" ] }, { "text": "Built an API over Azure Blob Storage that streams and analyzes 20+ GB files in place (no full download), producing cross-table data-quality reports in seconds and flagging source issues to fix at the source.", "project": "Azure Blob streaming API", "skills": [ "Azure Blob Storage", "Streaming", "Python", "API design", "Data quality", "Internal tooling", "Automation" ] }, { "text": "Automated recurring data-engineering work against the Snowflake API at scale: zero-copy clones for source-specific schemas, parallel test/data extraction, and Python-threaded data manipulation with structured logging.", "project": "Snowflake API automation", "skills": [ "Snowflake API", "Snowflake", "Zero-copy clones", "Threading", "Python", "Automation", "Internal tooling" ] }, { "text": "Designed and implemented an automated documentation system that extracts and structures raw-level metadata from multi-format, multi-language source files into table-based documentation across three languages.", "project": "multi-language documentation automation", "skills": [ "Metadata automation", "Automation", "Python", "JSON", "CSV", "XLSX", "Data modeling", "OpenAI API", "Azure ADP", "Documentation" ] }, { "text": "Co-designed and implemented reusable dbt macros for automated table and view generation, standardizing transformations across schemas.", "project": "dbt macro library", "skills": [ "dbt", "Automation", "Data modeling", "SQL" ] }, { "text": "Operate dbt Cloud as the team's orchestration layer: scheduled jobs, ad-hoc runs, and CI/CD pipelines for development and production deployments.", "project": "dbt Cloud operations", "skills": [ "dbt", "dbt Cloud", "CI/CD (GitHub Actions)", "Automation", "Internal tooling" ] }, { "text": "Got Claude Code adopted across the Data Office: built working demos for data engineering and data science, and authored an end-to-end setup guide (permissions, configuration, VS Code integration) now used by engineers in multiple Allianz country offices.", "project": "AI tooling adoption", "skills": [ "Claude Code", "AI-assisted development", "LLM API chaining", "Internal tooling", "Documentation", "Clear cross-functional communication" ] }, { "text": "Helped peer technical leads on other teams build their own agents on Allianz's internal AI platform and adopt Snowflake-API automation patterns for their projects.", "project": "cross-team enablement", "skills": [ "Mentorship", "Clear cross-functional communication", "Cross-team coordination", "AI-assisted development", "Snowflake API", "Internal tooling" ] } ] }, { "title": "Data Engineer Intern", "timeframe": "Feb 2025 - Jun 2025", "bullets": [ { "text": "Proposed and implemented a new data processing architecture that reduced end-to-end pipeline runtime by ~99 percent, cutting execution time from ~1.5 days to minutes.", "project": "pipeline architecture redesign", "skills": [ "ETL pipelines", "Batch processing", "Data modeling" ] }, { "text": "Rebuilt undocumented Oracle SQL pipelines into dbt models on Snowflake, achieving >=99.9 percent data accuracy, with ~70 percent of tables reaching full parity across millions of rows.", "project": "Oracle to dbt migration", "skills": [ "Oracle SQL", "dbt", "Snowflake", "Data validation", "Parity testing", "Data quality", "Large-scale migrations" ] }, { "text": "Simplified the data architecture by removing redundant processing steps, reducing compute costs and maintenance overhead.", "project": "pipeline optimization", "skills": [ "Data modeling", "ETL pipelines", "Batch processing" ] }, { "text": "Built a Power BI dashboard to monitor dbt Cloud runs with hierarchical filtering and enriched metadata for improved operational visibility.", "project": "dbt Cloud monitoring", "skills": [ "Power BI", "dbt Cloud", "Metadata automation", "Data quality" ] } ] } ] }, { "company": "NeoCru", "location": "Founder Project", "timeframe": "Apr 2025 - Dec 2025", "roles": [ { "title": "Founder & Full-Stack Engineer", "timeframe": "Apr 2025 - Dec 2025", "bullets": [ { "text": "Designed and built an AI-assisted recruitment platform enabling Dutch businesses to manage job postings and applications without full HR systems.", "project": "NeoCru platform", "skills": [ "Python", "Flask", "PostgreSQL", "API design", "Automation", "OpenAI API", "Data modeling" ] }, { "text": "Implemented an end-to-end full-stack system with a focus on data modeling, application workflows, and recruiter-facing dashboards.", "project": "NeoCru platform", "skills": [ "Python", "JavaScript", "Flask", "PostgreSQL", "API design", "Internal tooling", "Data modeling" ] }, { "text": "Integrated AI-driven automation to generate job descriptions and structure candidate data, reducing manual effort for recruiters.", "project": "NeoCru automation", "skills": [ "OpenAI API", "Automation", "NLP" ] }, { "text": "Designed secure, role-based access and dashboards for recruiters, focusing on data integrity, privacy, and maintainability.", "project": "NeoCru access control", "skills": [ "API design", "Data modeling", "Internal tooling", "GCP reCAPTCHA" ] }, { "text": "Deployed and operated the platform on Google Cloud Platform, using containerized services, CI/CD, and Cloud Storage for applicant resumes.", "project": "NeoCru deployment", "skills": [ "GCP Cloud Run", "GCP Cloud Build", "GCP Cloud Storage", "GCP Logs", "CI/CD (GitHub Actions)", "Containerization" ] } ] } ] }, { "company": "LinkInLead", "location": "Founder Project", "timeframe": "Jan 2024 - Dec 2024", "roles": [ { "title": "Founder & Full-Stack Engineer", "timeframe": "Jan 2024 - Dec 2024", "bullets": [ { "text": "Built an AI-driven automation platform for generating and publishing LinkedIn content, integrating the LinkedIn API with AI-based text generation.", "project": "LinkInLead platform", "skills": [ "Python", "Flask", "PostgreSQL", "OpenAI API", "Automation", "API design", "NLP" ] }, { "text": "Designed backend workflows for scheduled content creation, background processing, and job orchestration, emphasizing reliability and scalability.", "project": "LinkInLead workflows", "skills": [ "Python", "Automation", "API design", "Batch processing", "GCP Cloud Functions", "GCP Cloud Tasks" ] }, { "text": "Implemented secure data handling and session management for user accounts and publishing workflows.", "project": "LinkInLead security", "skills": [ "Python", "API design", "Data modeling", "Internal tooling", "PostgreSQL" ] }, { "text": "Deployed the system using containerized services and CI/CD pipelines, enabling automated updates and scalable execution.", "project": "LinkInLead deployment", "skills": [ "Containerization", "CI/CD (GitHub Actions)", "GCP Cloud Run", "GCP Cloud Storage", "GCP Logs" ] }, { "text": "Gained hands-on experience designing API-driven systems that coordinate external services, background jobs, and persistent storage.", "project": "LinkInLead architecture", "skills": [ "API design", "Automation", "Internal tooling" ] } ] } ] } ], "education": [ { "degree": "B.Sc. Cognitive Science and Artificial Intelligence (graduated)", "institution": "Tilburg University, Netherlands", "timeframe": "2022 - 2025" }, { "degree": "B.Sc. Engineering Science", "institution": "Tehran University, Iran", "timeframe": "2020 - 2022" }, { "degree": "Diploma in Mathematics", "institution": "Sanandaj NODET High School (National Organization for Development of Exceptional Talents), Iran", "timeframe": "2017 - 2020" } ], "projects": [ { "name": "Half Stupid - Teaching Minecraft Agents to Survive From Scratch", "timeframe": "Apr 2025", "bullets": [ { "text": "Built a dual-timescale neural network (reflex layer at every tick, context layer every 5 ticks) using shared learned embeddings across 2048 block/item/entity types, processing 557 raw inputs into 23 actions with zero human-labeled knowledge.", "skills": [ "Neural networks", "Python", "NumPy", "Feature engineering" ] }, { "text": "Implemented vanilla REINFORCE policy gradient training with survival-only reward (+1 alive, -10,000 on death), achieving emergent food-seeking behavior, distinct agent personalities, and inter-agent social dynamics without any reward shaping.", "skills": [ "Python", "PyTorch", "NumPy" ] }, { "text": "Profiled the Minecraft/Malmo substrate's scaling limits (4-agent ceiling, ~4,000 ticks/sec) and quantified the throughput ceiling capping further training experiments.", "skills": [ "Analytical mindset", "Python" ] } ] }, { "name": "Thesis - Predicting Perceptions of Dutch Company Names", "timeframe": "Jan 2025 - May 2025", "bullets": [ { "text": "Built an end-to-end data processing and modeling pipeline to predict human trait judgments (femininity, evilness, trustworthiness, smartness) from Dutch brand-like names.", "skills": [ "Feature engineering", "NLP", "Python", "pandas", "scikit-learn" ] }, { "text": "Transformed raw experimental ranking data into model-ready datasets by converting Parquet to CSV, translating metadata, and aggregating Bayesian posterior samples into stable regression targets.", "skills": [ "Parquet", "CSV", "Metadata automation", "Python", "pandas" ] }, { "text": "Engineered multiple feature pipelines, including character-level unigrams (27D), padded bigrams (479D), and contextual semantic embeddings using RobBERT (768D).", "skills": [ "Feature engineering", "RobBERT", "Transformers", "NLP", "Python" ] }, { "text": "Designed controlled experimental data splits to evaluate in-domain, out-of-domain, and few-shot generalization using fixed random seeds for reproducibility.", "skills": [ "Analytical mindset", "Feature engineering", "Python" ] }, { "text": "Trained and evaluated Elastic Net regression and feedforward neural networks, resulting in 64+ models with systematic hyperparameter tuning and consistent evaluation using R2.", "skills": [ "Elastic Net regression", "Neural networks", "scikit-learn", "Python" ] }, { "text": "Implemented the full workflow in Python, using pandas, numpy, scikit-learn, PyTorch, and Hugging Face Transformers, with version-controlled and reproducible experiments.", "skills": [ "Python", "pandas", "scikit-learn", "Transformers", "Git", "GitHub" ] } ] }, { "name": "Group Thesis - Multimodal Speech Recognition with AV-HuBERT", "timeframe": "Jan 2025 - May 2025", "bullets": [ { "text": "Evaluated audio-visual speech recognition using AV-HuBERT on the GLips German lipreading dataset.", "skills": [ "Transformers", "Computer vision", "Python" ] }, { "text": "Built an end-to-end preprocessing and alignment pipeline for lip-only video input, including lip patch extraction and audio-video synchronization using Dlib, OpenCV, FFmpeg, and Python multiprocessing.", "skills": [ "Dlib", "Computer vision", "Python" ] }, { "text": "Implemented and tested multiple inference configurations (audio-only, video-only, and audio-video fusion) using Hugging Face transformer APIs.", "skills": [ "Transformers", "Python" ] }, { "text": "Developed a parallelized inference and evaluation setup to optimize GPU utilization and automatically match and process 21,000+ audio and video files.", "skills": [ "Automation", "Python" ] } ] }, { "name": "Software Engineering Course - PetMatters", "timeframe": "Aug 2024 - Dec 2024", "bullets": [ { "text": "Led a team of 8 members, coordinating task allocation, technical decisions, and delivery for a semester-long AI-powered application.", "skills": [ "Clear cross-functional communication", "Analytical mindset" ] }, { "text": "Designed a multi-service architecture with separate orchestration, AI, and interface components, focusing on modularity, isolation of concerns, and scalability.", "skills": [ "API design", "Internal tooling", "Automation" ] }, { "text": "Trained and integrated T5-based language models for English-to-Dutch translation, text enhancement, and form-driven content generation using a university GPU server.", "skills": [ "T5", "Transformers", "NLP", "Python" ] }, { "text": "Defined service boundaries and data flow between components, gaining hands-on experience with distributed system design in an academic setting.", "skills": [ "API design", "Analytical mindset" ] } ] }, { "name": "C++ Course Project - Battle-C", "timeframe": "Aug 2024 - Dec 2024", "bullets": [ { "text": "Designed and implemented a modular C++ application with clear separation of game logic, state management, and AI behavior.", "skills": [ "C++", "Internal tooling" ] }, { "text": "Developed AI-driven bot logic to simulate opponent decision-making within a turn-based game system.", "skills": [ "C++", "Automation" ] }, { "text": "Implemented real-time game state tracking, including scores and entity status, reinforcing structured program design and state consistency.", "skills": [ "C++", "Data modeling" ] } ] }, { "name": "AI for Nature and Environment Project", "timeframe": "Aug 2024 - Dec 2024", "bullets": [ { "text": "Collected, cleaned, and processed environmental and fire occurrence data for predictive analysis.", "skills": [ "ETL pipelines", "Data modeling", "pandas", "Python" ] }, { "text": "Built a LightGBM-based prediction pipeline to model fire outbreak risk using environmental variables.", "skills": [ "LightGBM", "Feature engineering", "pandas", "Python" ] } ] } ] }