hrishikesh ~ zsh
~ whoami
Hrishikesh Saraiya
~ cat role.txt
~ cat status.txt
Software Engineer @ Upstart Arlington, VA
scroll

About

Systems Thinker, Life long learner, Type 1 Diabetic

I was diagnosed with Type 1 Diabetes at 12, which shaped how I think about what it means to break my life into systems and rely on technology to live. I found engineering through FRC robotics in high school, studied CS and Statistics at Ohio State, and have spent my career on ML platform teams at Capital One and now Upstart, focused on MLOps, Kubernetes, and the infrastructure that makes machine learning actually work in production.

Outside of work, I am an avid tennis fan who goes to the US Open every year, a die-hard Arsenal supporter, and an Ohio State football fan. I read widely — history, geopolitics, and investing. I'm a vegetarian who loves cooking at home and trying new recipes. I travel to India often to visit family, and I'm a practicing Hindu with a deep interest in Vedanta philosophy. I try to see everything in life through a systems lens.

Tools & Technologies

Python Kubernetes Docker AWS Kubeflow Metaflow Systems Design
T1D 14 years with T1D
Ohio State
CS + Statistics
Upstart
ML Platform Engineer
Arsenal
COYG

Experience

Where I've worked

Upstart

Software Engineer, ML Platform

Upstart
Dec 2025 — Present · Arlington, VA (Remote)

Upstart is an AI lending platform that uses machine learning to go beyond the FICO score in assessing credit risk. I sit on the ML Platform team, building the infrastructure and tooling that enables data scientists to train, validate, simulate, and deploy the models powering Upstart's borrower application process.

Python Kubernetes Metaflow
Capital One

Software Engineer, ML Platforms

Capital One
Aug 2021 — Nov 2025 · McLean, VA
Enterprise Model Platform

Built platform capabilities powering the full ML lifecycle — training, pipelines, deployment, and reusable SDKs — used by hundreds of data scientists across the company. Was a core member of the air-gapped pilot deployment of Weights & Biases on Capital One's Kubernetes infrastructure, driving enterprise adoption at launch. Developed a Python ML component framework that decoupled ML code from orchestration engines, becoming the standard for Kubeflow Pipelines component development company-wide. Core contributor to SDK standards on Capital One's Generative AI Platform.

Data Insights Platform

MLOps and SRE for Capital One Data Insights, an enterprise ML workflow orchestration platform. Maintained a Kubernetes fleet on AWS supporting data scientists' ML workflow jobs — infrastructure that cumulatively saves the enterprise $100M+ annually. Contributed to a mirror fleet for a commercial offering of the platform. Improved observability through a Python logging framework with OpenTelemetry and New Relic trace injection, and added CPU/GPU and memory monitoring for high-impact workloads.

Python Docker Kubernetes AWS Weights & Biases Kubeflow OpenTelemetry New Relic SRE MLOps

Projects

Things I've built

Dexcom G7 macOS Menubar App

A lightweight macOS menubar app for monitoring Dexcom G7 CGM readings in real-time. Displays your current glucose value and trend arrow directly in the menubar with color-coded status indicators — green for in-range, red for low, orange/yellow for high. Includes smart notifications for low and high alerts (with a 15-minute cooldown to avoid spam), a recent history dropdown showing the last 12 readings, and automatic updates every 5 minutes. Credentials are stored securely in the macOS Keychain.

  • Color-coded glucose status with trend arrows (⬆⬆ → ⬇⬇)
  • Smart alerts for lows and highs with throttled notifications
  • Auto-refreshes every 5 minutes, manual refresh on demand
  • Secure credential storage via macOS Keychain
Python macOS Dexcom Share API macOS Keychain

more coming soon — follow along on GitHub

Blog

Latest writing

View all

Thoughts incoming — check back soon.