Featured System
Distributed MapReduce Framework
C++ MapReduce engine deployed on EC2 with custom shuffle and performance tuning.
C++gRPCProtobufConcurrency+2
Pratyush Kumar – Distributed systems engineer portfolio
Formerly at Enphase Energy · MSCS @ Penn State
Pratyush Kumar
Backend Engineer
Performance Engineer
I build software systems that are fast, reliable, and built to hold up under real-world use.
Get in TouchAbout
I turn ambiguous backend problems into reliable, maintainable systems.
I'm Pratyush, a software engineer who takes ownership from problem definition through production. I care about clear abstractions, thoughtful tradeoffs, and software that remains understandable as it scales.
Ownership
I take work from an unclear problem to a shipped, measurable result and stay accountable for how it behaves in production.
Engineering judgment
I use data to find the real constraint, weigh tradeoffs, and choose solutions that fit the system instead of chasing complexity.
Reliable execution
I build with testing, observability, correctness, and maintainability in mind—not as cleanup after the feature ships.
I'm currently pursuing an MS in Computer Science at Penn State, with a focus on backend systems and distributed systems.
Snapshot
5,500+
users supported through production IoT systems
25%
fewer integration failures at Enphase
7x
throughput improvement in C++ systems work
MSCS
focus on backend systems and distributed systems
Recent highlights
- Built Holocron Timeline Engine with FastAPI, Next.js, and Neo4j for causal traversal and world-state reconstruction.
- Built FairShare as a ledger-first expense tracker with deterministic money handling and explainable balances.
- Benchmarking and optimizing GEMM workloads across CPU and GPU architectures for systems performance analysis.
- Pursuing an MS in Computer Science at Penn State University with a focus on backend systems and distributed systems.
Projects
Selected systems, products, and team collaborations I’ve built or contributed to.
Featured Product
FairShare
Ledger-first expense tracker built for explainable balances, deterministic money handling, and safe retries.
JavaSpring BootPostgreSQLNext.js+2
More Projects
Holocron Timeline Engine
Graph-backed timeline system for causal traversal, break simulation, and world-state reconstruction.
FastAPINext.jsNeo4j+3
Faithful Clinical Summarization
Clinical summarization pipeline that evaluates generated discharge notes through atomic claim verification.
PythonPyTorchQLoRA+3
Communication-Avoiding GEMM
CPU and GPU matrix-multiplication study spanning cache-aware OpenMP, tiled CUDA execution, cuBLAS, and roofline analysis.
CUDAC++cuBLAS+3
Distributed Key-Value Store (ABD)
Quorum-based key-value store comparing linearizable ABD against a blocking baseline.
C++gRPCProtobuf+2
App-Aware Clipboard Manager
Privacy-first macOS clipboard manager designed around correctness instead of feature bloat.
PythonPyQt6SQLite+2
InstructBLIP Image Captioning
Vision-language experimentation toolkit for zero-shot evaluation and parameter-efficient Flickr8k fine-tuning.
PythonInstructBLIPQLoRA+3
AR Object Rendering Pipeline
Computer-vision pipeline that reconstructs a scene, finds a dominant plane, and renders a 3D object into video.
PythonOpenCVCOLMAP+3
Education & Experience
Where I've worked, studied, and what I shipped.
Education
Aug 2025 — May 2027
The Pennsylvania State University — MS in Computer Science
State College, PA
GPA: 3.7 / 4.0Focus: Distributed systems, Computer Vision, Deep Learning, and High-Performance Computing
Built depth across operating systems, computer architecture, and concurrent matrix computation with an emphasis on performance and hardware-aware systems design.
Expanded into computer vision and NLP through graduate coursework spanning vision-language models, deep learning for language, and advanced computer vision.
Applied that coursework in systems-heavy projects including a distributed MapReduce framework and a quorum-based key-value store.
Built depth across operating systems, computer architecture, and concurrent matrix computation with an emphasis on performance and hardware-aware systems design.
Expanded into computer vision and NLP through graduate coursework spanning vision-language models, deep learning for language, and advanced computer vision.
Applied that coursework in systems-heavy projects including a distributed MapReduce framework and a quorum-based key-value store.
Jul 2019 — May 2023
Vellore Institute of Technology — BTech in Computer Science and Engineering
Vellore, India
GPA: 3.5 / 4.0Focus: systems, networks, cryptography, and core computer science fundamentals
Built a capstone around selective region and partial face-image encryption.
Coursework included algorithms, operating systems, database systems, computer networks, and data structures.
Built a capstone around selective region and partial face-image encryption.
Coursework included algorithms, operating systems, database systems, computer networks, and data structures.
Experience
Aug 2025 — Present
Penn State — Graduate Learning Assistant
State College, PA
Improved algorithmic proficiency for 60+ students by facilitating weekly lab sessions and real-time debugging support in C++ and MATLAB.
Led guided problem-solving sessions that helped students debug faster and build stronger systems-programming intuition.
Provided structured feedback during labs and office hours to improve student confidence and assignment quality.
Led guided problem-solving sessions that helped students debug faster and build stronger systems-programming intuition.
Provided structured feedback during labs and office hours to improve student confidence and assignment quality.
Jan 2023 — Jul 2025
Enphase Energy — Software Engineer
Bangalore, India
Reduced production integration failures by 25% by improving microservice reliability and applying TDD for systems serving 5,500+ daily users.
Optimized API performance and SQL query efficiency by fixing bottlenecks surfaced through DataDog monitoring, reducing database load.
Decreased manual regression testing by 50% by integrating quality gates into Jenkins CI/CD and tightening engineering standards.
Reduced onboarding ramp-up time by 65% by automating legacy workflows and mentoring new engineers on production systems.
Contributed to global compliance features that helped support product expansion into European markets.
JavaRuby on RailsPostgreSQLJenkinsSonarQubeDataDogRSpec
Optimized API performance and SQL query efficiency by fixing bottlenecks surfaced through DataDog monitoring, reducing database load.
Decreased manual regression testing by 50% by integrating quality gates into Jenkins CI/CD and tightening engineering standards.
Reduced onboarding ramp-up time by 65% by automating legacy workflows and mentoring new engineers on production systems.
Contributed to global compliance features that helped support product expansion into European markets.
JavaRuby on RailsPostgreSQLJenkinsSonarQubeDataDogRSpec
Aug 2021 — Jan 2022
Acies Consulting — Full-Stack Developer Intern
Mumbai, India
Built workflow automation and data tooling that accelerated financial operations and improved backend interoperability.
Accelerated financial workflow setup from 3 days to 30 minutes by building interactive JavaScript and AJAX components for data operations.
Architected pluggable backend connectors so users could route workloads across MySQL, PostgreSQL, and Redis targets.
Automated stock-metric documentation with a Python and BeautifulSoup scraping pipeline for Yahoo Finance data.
DjangoJavaScriptAJAXPythonMySQLPostgreSQLRedisBeautifulSoup
Accelerated financial workflow setup from 3 days to 30 minutes by building interactive JavaScript and AJAX components for data operations.
Architected pluggable backend connectors so users could route workloads across MySQL, PostgreSQL, and Redis targets.
Automated stock-metric documentation with a Python and BeautifulSoup scraping pipeline for Yahoo Finance data.
DjangoJavaScriptAJAXPythonMySQLPostgreSQLRedisBeautifulSoup
Engineering Toolkit
The languages, frameworks, and delivery tools I use to build backend systems, performance-sensitive software, and infrastructure-heavy products.
Backend-firstProduction-testedSystems-minded
Systems & Core Languages
The foundation I use for performance-sensitive systems, APIs, and application logic.
C++
Python
Java
TypeScript
JavaScript
SQL
Multithreading
Backend & Distributed Systems
The stack behind production services, concurrency-heavy systems, and networked applications.
gRPC
Protobuf
REST APIs
AWS (EC2)
Concurrency
Distributed Systems
Quorum Protocols
Redis
PostgreSQL
MySQL
Frameworks & Product Engineering
The frameworks I reach for when shipping end-to-end products with real users and operational constraints.
Ruby on Rails
Spring Boot
FastAPI
Django
React
Next.js
Vite
PyQt6
Docker
Flyway
Supabase
Frontend & UI
Enough frontend depth to build interfaces that feel intentional, clear, and usable.
React
Next.js
Vite
Tailwind CSS
HTML
CSS
JavaScript
Delivery, Testing & Observability
The tooling I rely on to ship safely, monitor behavior, and improve reliability over time.
Git
Linux
Jenkins
pytest
RSpec
Selenium
DataDog
SonarQube
Unit & Integration Testing
Performance Benchmarking
Profiling
Data & Persistence
Storage and data tools used across product platforms, automation, and internal workflows.
PostgreSQL
MySQL
Redis
MongoDB
Neo4j
Contact
Let's talk about systems, products, and hard backend problems.
If you're building something real and need a steady engineering partner, I'd like to hear about it.
Resume available on request.