Engineer — San Francisco, CA

Building
things
that help
us connect.

Back-end engineer with 5+ years at Microsoft and Kwartile. Now pursuing robotics, real-time sensing, and human-centered systems.

View Selected Work
01

About

Me

I'm a back-end engineer with 5+ years building scalable, cloud-native systems. From real-time data pipelines at Microsoft to ML-driven platforms at early-stage startups.

After earning a Computer Science degree from UC Davis, I joined a startup where I built an entity resolution model with 97% accuracy, spun up real-time streaming infrastructure, and engineered a data platform from concept to production. Then I scaled up at Microsoft, working on critical tasks like reducing API latency by 59% and automating 80% of onboarding workflows.

I took a deliberate sabbatical in 2024 to redirect. I discovered a home for my skills in robotics, started teaching myself ROS2, building with simulation tools, participating in a hackathon, and built my own arm end-to-end. Now, I am actively looking for projects where my industry skills in Software Engineering can be useful to the Robotics space.

0
Years industry experience
0
API latency reduction at Microsoft
0
Hours of manual work automated
0
ML model accuracy
02

Selected Work

SimulationInverse KinematicsMuJoCo

Inverse Kinematics Robotic Arm

A fully working pick-and-drop simulation built from fundamentals in MuJoCo and Python. Based on Jacobian calculation, damped least squares, and IK tuning.

View project
RoboticsROS2Human-Robot Interaction

Dice-Rolling Robot

A physical robot designed from scratch to respond to human interaction: rolling dice on command. Built to learn the fundamentals of ROS2, Gazebo, and RViz through hands-on construction.

View project
Team Photo
HackathonRoboticsIntersectional Design

Sauceathon: Robotics Hackathon

A 24-hour robotics hackathon where I collaborated with a mixed software and mechanical engineering team to build a robot designed to interact with both humans and other robots. A lesson in intersectional design and diverse perspectives.

View project
Platform Photo
ML PlatformPrivacy EngineeringPython

Datamorph.AI

A data platform that makes it easy to incorporate privacy-safe practices into ML pipeline building — applying sanitization and vector embedding with industry best practices to protect personal information.

View project
EMG signal Noisy · user-specific · context-dependent Feature extraction Combined movements, not discrete gestures Training data Patients + control group Cascaded ML models Generic model Broad population baseline Personal model Adapts to individual user Homomorphic encryption Edge device Smartphone · low latency Motor intent → action Real-time output · reliable actuation Prosthetic limb Co-robot · negotiates intent & trust Continuous recalibration Goal: 100% functionality return Accessible · private · accurate
Research ProposalAssistive RoboticsDistributed Computing

Intelligent Prosthetics

An ongoing research proposal: designing an accessible, private, and accurate robotic prosthetic that returns full user functionality and offers personalized collaboration during rehabilitation. Driven by distributed computing and ML.

Read proposal
03

Experience

Oct 2024 — Present
Personal Sabbatical
Independent
  • Self-teaching robotics fundamentals: ROS2, Gazebo, RViz
  • Built a dice-rolling robot; participated in the Sauceathon hackathon
  • Directed creative and research projects aligned with robotics & human-centered AI
Nov 2021 — Sep 2024
Software Development Engineer II
Microsoft — Data & AI Platform
  • Reduced API latency by 59% across four production endpoints via Azure API Management
  • Automated 80% of manual onboarding effort — 960+ hours saved with Azure Durable Functions
  • Cut developer request time by 60% and reduced errors 15% via Logic App workflows
  • Increased service reliability coverage by 30% through real-time monitoring improvements
Mar 2019 — Jun 2021
Software Development Engineer
Kwartile
  • Built entity resolution ML model with 97% accuracy using Spark, SQL, and HDFS
  • Architected ETL platform APIs integrating Databricks, AWS Glue, and Spark-Standalone
  • Delivered real-time streaming infrastructure for Nordstrom — deployed to 5 teams
04

Skills

Languages

  • Java
  • Python
  • C#
  • SQL

Back-End & Data

  • Apache Spark / Delta Lake
  • Kafka & Redis
  • Spring Boot
  • REST APIs

Cloud & DevOps

  • Azure / AWS
  • Databricks
  • Docker & Kubernetes
  • CI/CD Pipelines

Robotics (Learning)

  • ROS2
  • Gazebo / MuJuCo
  • RViz

ML & Data Science

  • Entity Resolution
  • Feature Engineering
  • Vector Embeddings

Methods

  • Distributed Systems
  • Real-Time Pipelines
  • Platform Architecture

Let's
work
together.

Emailparushi@gmail.com
Phone(925) 353-0016
LinkedInLinkedIn Profile →
Based inSan Francisco, CA