Skip to content

Devank Uppal

Data engineer

  • Green card holder · no sponsorship needed
  • Open to full-time roles from May 2027
  • B.S. Computer Science, Iowa State University · May 2027
  • Based in Ames, IA

Computer Science senior at Iowa State. I build pipelines that turn raw data into answers: AWS Glue, S3 and Athena automation at Spectrum, PulseIQ data ingestion at Fiakominds, and PySpark and Databricks on machine telemetry at John Deere.

Portrait of Devank Uppal

Experience

  1. Data Analyst (part-time, during school) · John Deere

    Current

    Aug 2026 – present · Intelligent Solutions Group · Ames, IA

    data lake my scripts join telemetry and engineering sources from
    PB-scaledata lake my scripts join telemetry and engineering sources from
    on Databricks, from raw machine data to analysis-ready views
    PySpark + SQLon Databricks, from raw machine data to analysis-ready views
    • Write PySpark and SQL on Databricks that join machine telemetry and engineering sources from a petabyte-scale data lake into analysis-ready views.
    • Turn raw telemetry into reusable data views and Tableau dashboards for reliability analysis on new and existing products.
    • Support Right to Be Forgotten data compliance.
    PySparkSQLDatabricksTableau
  2. Software Developer (part-time, during school) · Fiakominds

    Current

    Apr 2026 – present · St. Louis, MO

    building the data ingestion for the customer sentiment product
    PulseIQbuilding the data ingestion for the customer sentiment product
    account created, license assigned, login details sent, with no one in the loop
    Automaticaccount created, license assigned, login details sent, with no one in the loop
    • Building the data ingestion for PulseIQ, the company's customer sentiment product.
    • Built the onboarding platform's data flow on Supabase (Postgres): account creation and license assignment trigger the login-details email automatically, a step someone used to do by hand.
    Supabase (Postgres)Data ingestionTypeScriptReactfiakominds.ai
  3. Advanced Technology Intern (summer) · Spectrum

    May – Jul 2026 · 28-person migration team · St. Louis, MO

    In production

    records extracted per run (154K accounts, 310K opportunities)
    464K+records extracted per run (154K accounts, 310K opportunities)
    match with the manual export it replaced
    100%match with the manual export it replaced
    per run, was 45 min to several hours (~2 hrs of manual work)
    17 minper run, was 45 min to several hours (~2 hrs of manual work)
    • Automated the Salesforce extraction at the start of the migration pipeline: AWS Glue calls the Salesforce API and lands 464K+ records per run in S3, validated 100% against the manual export, in ~17 min instead of 45 min to several hours. In production.
    • Built a contract-database extraction that runs in parallel, on a dedicated service account with credentials in AWS Secrets Manager (Terraform), writing to a Glue table (built and tested; not in production when I left).
    • Built the post-migration Salesforce flag update: it computes which accounts need the flag and batch-updates only those through the Bulk API, with an approval step (built and tested; not in production when I left).
    • Coordinated project management for the 28-person migration team: leadership status reports, cascading timelines, and deployment handoffs between the business, the vendor and DevOps.
    AWS GlueS3AthenaSalesforce APISecrets ManagerTerraformPython

    Before · manual, waits on people

    1. Salesforce + contract DB

      the data the migration needs

    2. Manual export

      one person, by hand · 45 min to hours (~2 hrs of work)

      in parallel

      Manual query

      depended on one person's access

    3. CSV upload to S3

      by hand

    4. ETL & migration

      automated

    5. Account flags

      updated one account at a time

    • Blocked if one person was out
    • Unpredictable timing
    • Error-prone CSV handling
    • No audit trail

    After · automated

    1. Salesforce extract

      Glue + Salesforce API · ~17 min

      in production

      in parallel

      Contract DB extract

      service account · Secrets Manager

      built

    2. Amazon S3

      both extracts land here

    3. ETL & migration

      raw → staged → migrated

    4. Approval gate

      email approve / reject

      built

    5. Flag update

      Bulk API · thousands at once

      built & tested

    Pipeline monitoring web app

    built & tested

    Built to replace the command-line tool, so anyone on the team could run and watch the pipeline from a browser.

    1. Front end

      Web dashboard

      served from EC2

      • Start a run in one click
      • Live status for every step
      • Approve or reject a gate
    2. Backend

      Python API

      EC2, inside the VPC

      • Triggers the Glue jobs
      • Tracks each run's state
      • Emails approve / reject links
      • Writes the audit trail
    3. AWS

      The pipeline

      Glue · S3 · Secrets Manager

      • Extracts and migration jobs
      • Credentials never leave Secrets Manager
    • VPC inbound / outbound rules
    • Security groups
    • IAM roles
    • Set up with DevOps

Projects & research

Privacy-preserving analytics

Research assistant, Iowa State · Aug 2025 – present

Extended Google's RAPPOR into Multi-View RAPPOR so private aggregate statistics work on correlated data, and benchmarked Private Set Intersection for runtime, memory, accuracy and scale. I wrote the implementation.

PythonDifferential privacyPSI

HABET balloon telemetry

Used on 8+ launches

Iowa State · Make to Innovate · Jan – Dec 2025

A real-time telemetry pipeline: radio packets read over USB serial, parsed by a Flask data broker and published over WebSocket to any subscriber on our network, including the live dashboard and the rooftop antenna tracker (I implemented the tracking math a teammate designed).

PythonFlaskWebSocketUSB serial

Hospital Management System

1st-place team · 300-student cohort

Iowa State team project · Aug – Dec 2025

Role-aware data access for a hospital system: MySQL behind Spring Boot APIs, with a Gemini assistant that only sees what each role is allowed to.

JavaSpring BootMySQLGemini

OpenVINO device switching

Personal project · 2026

Real-time face detection that moves inference between CPU and GPU based on latency and system load.

PythonOpenVINO

Skills

Data
PySpark · Databricks · pandas · Tableau · ETL
Languages
Python · SQL · Java · TypeScript · C++
Cloud & DevOps
AWS (Glue, S3, Athena, EC2, Redshift, Secrets Manager) · Terraform · GitLab CI/CD · Docker · Linux
Web & backend
React · Flask · FastAPI · Spring Boot · REST APIs · WebSockets · Supabase (PostgreSQL) · MySQL

Education

B.S. Computer Science

Iowa State University · May 2027

Minors in Artificial Intelligence, Business and Statistics

Leadership

  • AWS Student Builder Campus Leader · AWS · Iowa State

    Apr 2026 – present · Outreach and classroom sessions that get students started on AWS.

  • Software Lead & Treasurer · VEX U Cyber Robotics

    Jan 2025 – present · Lead a 15+ person team on C++/PROS autonomy and driver control; own path planning, PID tuning and the budget.

अ

About the logo: dev reads as developer, and अंक (ank) is Hindi for number, so my name reads as software + data. Outside of work: cars, car culture and a Hot Wheels collection. Favorite car is the Nissan Skyline GT-R R34, which is where the dot in my logo comes from.

Contact

I’m looking for full-time software or data engineering roles starting after May 2027. Email is the fastest way to reach me.

uppal.devank@outlook.com

or leave a message