Dakota Parker Resume

719-433-0019 · dakotacolorado2@gmail.com · GitHub · LinkedIn

Summary

ML Systems Engineer with 4 years at Amazon building production inference infrastructure at Alexa scale. Specialized in inference optimization, model training and evaluation, and end-to-end ML pipeline development.

Skills

ML Systems & Serving — ONNX, Triton Inference Server, PyTorch, Amazon SageMaker (Endpoints, Training, Registry), Amazon Bedrock

Data Engineering & Pipelines — PySpark, Polars, Feature Engineering, Training Pipelines, DAG-based Orchestration, Online/Offline Eval

Cloud & Infrastructure — AWS (EMR, ECS, Batch, Glue, Athena), Infrastructure as Code (CDK, Terraform), Distributed Systems

Languages — Python, Java, TypeScript, SQL

Experience

Amazon — Artificial General Intelligence

Software Development Engineer II · 07/2024 – 07/2026

  • Architected the production infrastructure and ML Ops lifecycle for a 334M-parameter cross-encoder ranking model used in Amazon Local Search; served millions of realtime Alexa requests with sub-second latency and zero production incidents since its January 2025 launch.
  • Optimized ranking model runtime inference, reducing P50 inference latency by 88% (700ms to 80ms) and compute infrastructure costs by 90%; replaced monolithic P5 instances with a parallelized fan-out architecture on g4dn instances using ONNX runtime and client-side request batching.
  • Built a model release pipeline to automate offline model conversion, evaluation, and production release; reduced release lead time from 3 weeks to 3 days. Enabled 6 zero-rollback releases driving a 17% search quality lift across 2025 — the team's largest source of quality gains in the year.
  • Led replacement of an LLM-based entity resolution routine with a BERT cross-encoder, reducing monthly inference costs from $110K to $700 and increasing throughput by 1,000×; fine-tuned on 2M historical LLM labels and deployed on g4dn.
  • Built a shared Python evaluation library adopted by 12 engineers and scientists; standardized metric definitions across online and offline workflows, scaling from 1 to 15+ metrics in 3 months and replacing ad-hoc reporting with consistent, reproducible evaluation across the team.
  • Partnered with applied scientists to develop lightweight Python tooling and infrastructure to automate and orchestrate experimental offline workflows. On-boarded and automated 15 scientist-authored ETL jobs for feature enrichment, training data generation, and online evaluation.

Amazon Web Services

Software Development Engineer II · 01/2022 – 07/2024

  • Led migration planning and development to transition legacy Redshift workflows to a serverless Glue/Athena architecture; partnered with engineers across the org to migrate ingestion feeds for 30+ data sources, resulting in $360K/year in savings.
  • Automated AWS Control Tower service deployment using AWS CDK; developed a TypeScript library distributed across 5 services to standardize service deployment routines; compressed new region release cycles from 2 months to 1 week, directly enabling the launch of AWS Control Tower in 8 new global regions in 2023.
  • Supported the development and launch of a centralized AWS security compliance feature; owned security review, load testing, and integration testing for a distributed backend service managing 200K+ control instances and 1M+ weekly events.

LPL Financial

Senior Data Engineer · 08/2020 – 01/2022

  • Lead engineer for the "Enhanced Analytics" data product, driving annual contract growth from $1M to $6M in 2021; scaled a 3-customer pilot into a production system serving 34 financial institutions across 48 unique dataset configurations.
  • Designed a configuration-driven workflow and Python execution engine that replaced hardcoded SQL with Jinja2 templates; automated concurrent job execution and backfills, reducing runtime from 16 to 3 hours and enabling a 10x expansion in product offerings.
  • Received the LPL Extraordinary Achievement Award for engineering contributions that directly enabled a $5M expansion in annual data sales.

Education

University of Colorado, Boulder

B.S. in Applied Mathematics · GPA 3.80 · 08/2015 – 05/2018

Minors in Computer Science & Physics · College of Engineering and Applied Science

Stanford University

Additional Graduate Coursework · GPA 3.70 · 08/2024 – 12/2024

CS229: Machine Learning (A−) · Project: Chinese Checkers Machine Learning Model