Zain A.

I build data pipelines that pull data from APIs, filings, and databases, clean it, and load it into a warehouse your team can query and trust. I've worked as a data engineer for 3+ years, mainly with Python, SQL, Airflow, BigQuery, AWS, GCP and Snowflake. Most of my recent work is for finance and research clients who need data that is correct as well as fast. My focus is making sure the numbers are correct by the time they reach a report, because a fast pipeline with wrong data costs clients more than a slow one. WHAT I'VE BUILT: ✅ SEC filings pipeline for a finance research firm: - A daily script retrieves new filings from SEC EDGAR. It includes a backfill to 2021 for any fund added later. - Raw filings are stored in S3, then processed by a dedicated parser for each fund type. Auto ABS filings cover note classes and 30/60-day delinquency data. BDC and private equity filings cover the schedule of investments. - A validation step checks the parsed values and normalizes scales, since filings report numbers in different units, before loading into PostgreSQL. - The clean data feeds dashboards that update automatically and LLM-generated reports for the firm's clients. - Funds file quarterly or semi-quarterly, so the pipeline has to handle new layouts and late filings without breaking. ✅ LLM data pipelines for an SEC research intelligence platform: - These pipelines turn public filings into structured data that an AI research layer can search and reason over. - The work covers extraction, cleaning, and storage design, so the LLM receives consistent inputs. ✅ GCP data pipelines: - API data is ingested with Cloud Functions and scheduled with Cloud Scheduler. - Multi-step workflows are orchestrated in Cloud Composer (Airflow), with retries and failure alerts. - BigQuery warehouses use tables designed around how the business actually queries them. WHAT I CAN HELP WITH: - New ETL/ELT pipelines from APIs, databases, files, or public data sources - Moving scattered data (spreadsheets, CSV exports, separate databases) into one warehouse - Airflow setup, DAG design, and fixing unreliable scheduled jobs - BigQuery or Snowflake data modeling and query performance - Extracting structured data from documents and filings, including LLM-based extraction - Data validation, so reports and dashboards show the right numbers TOOLS I USE: - Languages: Python, SQL - Orchestration: Apache Airflow, Cloud Composer, Cloud Scheduler - Warehouses and databases: BigQuery, Snowflake, PostgreSQL - Cloud: GCP (Cloud Functions, Cloud Storage, BigQuery), AWS (S3) - LLM tools: OpenAI and Claude APIs, prompt design, structured JSON outputs, and Python for the full pipeline around them. - LLM and agents: OpenAI API, Anthropic Claude, LangChain, LangGraph, LlamaIndex, Pinecone, Sentence Transformers - Other: REST APIs, Git, pandas, Docker, LLM APIs for extraction, dashboarding tools I use AI coding tools like Claude Code and Cursor to work faster. I still check and test everything myself before I hand it over. While I work, I send you short updates so you always know where things stand. If something in the requirements is unclear, I ask right away so we can fix issues early. Have data in APIs, files, or spreadsheets that you want in one place? Send me a message with what you have and where you want it to go.