AI Data Migration Engineer

About the Role

Every new Gov Pilot customer arrives with years of data locked in legacy systems. Your job is to get that data into Gov Pilot (often called data conversion in government software) accurately and quickly, and to make every migration faster than the one before.

AI comes first in this role. You’ll direct AI agents such as Claude Code to profile legacy databases, draft field mappings, generate transformation and load scripts, and automate validation. Your value is in directing that work, verifying it, and turning it into repeatable tooling, not in writing every line yourself. If your instinct is to hand-code a solution from scratch, this isn’t the right role.

If your instinct is to get an AI to build it, prove it’s correct, and make it reusable, keep reading.

To be clear: you need to be technical enough to know when the AI is wrong. You’ll be reading and fixing generated SQL and scripts every day.

You’ll own your work end-to-end with little day-to-day direction. We’re looking for someone who sees a slow process and fixes it without waiting to be asked.

• Own migrations end-to-end. Take raw client data (database backups, SQL Server/Access/Oracle exports, CSV and Excel files) from intake through mapping, transformation, loading, and sign-off.
• Drive the work with AI. Use AI agents to profile source databases, propose source-to-target mappings, and generate cleansing, deduplication, and load scripts. Then review, test, and take ownership of the output.
• Build tools, not one-offs. Turn each migration into reusable, AI-assisted import pipelines, prompt and agent workflows, and scripts, so the next customer on the same legacy system takes less time.
• Automate validation. Replace manual spot-checks with automated reconciliation: row counts, checksums, and field-level diffs. The bar is zero data loss and field-level accuracy, proven rather than assumed.
• Push AI adoption forward. Find the next manual step in the migration process and automate it. Share what works with the team and help set how Gov Pilot runs migrations.
• Catch problems early. Find data quality issues soon enough that they never threaten a go-live date, and explain them clearly to the people who need to act.
• Keep it auditable. Keep mapping logic, scripts, and logs clear enough that any teammate can trace how a record was transformed.
• Be there for go-lives. Production cutovers often happen in evenings or on weekends to avoid disrupting town operations. You’ll plan for these and be available to see them through.

• 15 days: You’ve completed a migration alongside the team and mapped where the current process is manual.
• 30 days: You’re owning migrations independently and have automated at least one recurring manual step.
• 45 days: You’ve shipped a reusable import pipeline for a common legacy system and measurably cut time per migration.

• 3+ years of hands-on experience with relational databases and data imports : you’ve moved real data between real systems (ETL, bulk loads, data conversions, or migrations)
• Strong SQL: you can write, debug, and optimize queries, and you can quickly spot when SQL an AI generated is wrong
• Ability to read and modify scripts (Python, Power Shell, or similar) that an AI generated
• Daily, hands-on use of AI coding agents (e.g., Claude Code, Cursor, Copilot agents) to build data workflows. You can show us something you built this way and tell us what the AI got wrong.
• A track record of automating work that used to be manual , whether a process, a report, or an import
• Self-direction: you’ve owned projects with little supervision and delivered without being chased
• Willingness to work evenings and weekends when a cutover or go-live requires it
• Care with…