Navid Heydari

Navid Heydari

Principal Architect

Seattle, WA

About Me

A technology enthusiast with over a decade of experience working on cloud-native enterprise systems. I enjoy the challenge of AI/ML architecture, thinking through responsible AI, and contributing to distributed systems that scale — always with a lot still to learn along the way.

Experiences

Senior Tech Lead

2025 - Present · DocuSign

  • Designed and implemented a graph-based deep neural network with a high-performance data processing framework handling 25B+ records daily in under 20 minutes.
  • Applied smart sampling to handle an imbalanced, noisy dataset and scaling to reduce data volume, cut compute costs, and enable flexible execution across single-instance and distributed architectures on CPU/GPU.
  • Designed customizable clustering with explainability baselines to ensure transparency and trust in model outputs, aligning with Responsible AI principles.
  • Initiated and standardized on-call and incident-resolution processes.
  • Pioneered an Agentic RAG architecture for responsible autonomous agents, enabling automated self-healing and support for impacted customer accounts.

Senior Architect / AI Architect

2017 - 2025 · T-Mobile

  • Designed and implemented a prediction model based on highly imbalanced customer secure data to provide tailored marketing campaigns, reduce costs by 4000x, and increase revenue with a higher lift score than vendor product suite.
  • Designed ML pipeline and set of controls to onboard data scientists faster and deliver models to production safely and on time.
  • Extracted insights on customer segments (churn cohorts) and generated self-explanatory reasons and attributes to clarify each triggering factor.
  • Designed and implemented cloud-based back-end microservices for 99.99% availability and resiliency, decreasing launch configurations from 48 hours to less than an hour.
  • Designed customized priority-based event ingestion architecture.
  • Established and standardized SLA for microservices.
  • Led and implemented security architecture and standards in cloud-based ecosystems.

Software Engineer

2013 - 2016 · Fiserv

  • Developed a robust solution for processing and validating new MasterCard credit card numbers (2-series BIN).
  • Implemented performance tuning for internal processes (crunch 1 sec to 10 ms) for each credit card validation.
  • Designed and executed disaster recovery processes and critical test scenarios, restoring 52K+ bill notifications within an hour.
  • Designed complex set of customized external monitoring tools for different Billers, increasing processing speed of 100M+ records.
  • Achieved Team Lead role for domestic-sold product suite, creating enhancement requirements, upgrading, patch development, and maintenance of nationwide financial software.

Education

Master of Science (MS) in Computer Science and Systems

2021 · University of Washington

Thesis: Enhancing Observability of Serverless Computing with the Serverless Application Analytics Framework (2021)

Bachelor of Science (BS) in Computer Science

2010 · Baha'i Institute of Higher Education (BIHE)

Thesis: Exploring research profiling by analyzing techniques, building a knowledge tree, and mining research articles (NLP datasets) to identify emerging research trends

Publications

Skills

ML & AI — MLFlow (Databricks), Snowflakes, PySpark, PyTorch-Geometrics (GNN), TensorFlow, RAG-based design, LLM-based, Vector database (Chroma, PineCone, etc.), Time-series LSTM, Agentic Architecture

Languages — Java, Python

Cloud & DevOps — Cloud Foundry, K8s & Docker, AWS (ElasticSearch, Lambda, S3, CloudWatch,CloudFormation, SNS/SQS), Jenkins, Prometheus, Terraform, Azure AI-Studio, GCP Compute clickhouse, Azure Kusto

Frameworks & Tools — Spring Boot, Maven, Bash, Hadoop, Redis, RabbitMQ, Kafka, Apigee, CouchDB, Grafana, OpenWhisks|IBM Bluemix

Misc. — R, C(90 & 99), C++, Assembly(AVR $ 8086), Go, TypeScript, TCL, JavaScript Angular, MongoDB

Blog

Python Design Pattern - Factory and Singleton along with Callable Wrappers

Apr 10, 2026  · 6 min read

Developing a connection Factory Wrapper to holds the singleton opject for heavy Object creation…

High-Performance Observability Pipeline

Apr 13, 2025  · 3 min read

High-Performance Observability Pipeline for AWS | databricks and ML model stack.

How to Approach to System Design Questions

Apr 8, 2025  · 3 min read

Thinking checklist during System Design Problems

System Design: Large-Scale Sensor Data Ingestion

Dec 15, 2024  · 4 min read

Architecture for ingesting real-time temperature data from 1 million sensors, serving a live heatmap…

Graph Theory - Part 2 Add More Restrictions and Rules

Nov 20, 2024  · 4 min read

Graph theory p2 - Some Restrictions getting applied

All posts →