Harshal Parekh

Senior Software Developer based in Austin
Responsible for API / System Design of Athena at Visa
& backend development and UXR of Paradise IoT Dashboard at Digiotouch.
Master's — Carnegie Mellon.

Freelancing for small / medium projects.
work and projects

Selected Deep Dives

In all code, I strive for brevity (without loss of legibility).
The less code there is, the less places there are for bugs to lurk.


Automated payment gateway validation, and more

Working closely with multiple teams in the core payments department to develop tools for seemingly unsolvable problems and create experiences that connect design and human behaviour

Reduced gateway onboarding time-to-market by 30% — Athena
1MN internal-only hits on Minerva
Approved Visa Patent for Athena's Permutation's Engine
3 over-achievers awards in 2020

Project Lead, Innovation Wing // Sr. Software Developer
API & System Design // Full Stack & UX Research

Read  case study

All-in-one platform to view real-time data from IoT devices

Designing & Developing the complete dashboard from scratch while also delivering quality user experiences across a broad platform

SSO APIs for sibling companies
Self-healing dockerized micro-services on Google Kubernetes Engine
Zero downtime CI/CD pipelines
Real-time chart updates

Paradise IoT Dashboard // Project Lead

read case study

User Recommendation and Topic Words Extraction from 100MN Tweets

A scalable, reliable web-service over 100MN tweets on AWS
A 3 week cloud computing academic project extended and improved over the summer

Query 1: Is this a valid blockchain transaction? RPS: 60,000
Query 2: What friends to recommend to this user? RPS: 22,000
Query 3: What are the to N topic words in this country and during this time? RPS: 6,000
First on the leaderboard for the mixed query
Achieved maximum RPS under the budget of $0.85/hour on AWS

Backend Engineer // High performant web-services

Read case study

Processing stream data and a variety of ML models stitched together

Complete backend designing and development of a cab service by handling large stream of data in real-time, predicting cab prices with a high accuracy, and end-to-end backend accepting speech queries for rides and responding with a speech-based answer

Matched 10,000+ customers to drivers per second deployed on AWS EMR
Hyper-parameter tuning of XGBoost model predicting cab fares
GCP text-to-speech and speech-to-text
Cloud Vision & AutoML to identify locations from photos

Stream-Processing Engineer // Process Kafka messages
ML Engineer // XGBoost tuning & GCP ML Apps and Services

Read case study

Latest Articles

Some of my best pieces, learning experiences, opinions, and guides.

Branch Prediction — Everything you need to know

Why would processing a unsorted array take almost 3 times as the time taken to process a sorted array?

250,000 views on Medium

Make your résumé stand out

Your entire career is put on a piece of paper, don't make it look like a newspaper.

mostly viewed by new grad students

What is “yield” or “generator” in Python?

You shouldn't just load a huge file into memory, it is expensive, read more to understand why.

200,000 views on Medium

I no longer answer on Stackoverflow

There are more than 20 million questions on Stack Overflow; what are the chances yours isn't asked before?

mostly viewed by ... idk, me?