Open to Summer 2027 SWE & DS internships
Hi, I'm Justin.
Distributed systems engineer building high‑performance backend & data infrastructure
Go
Python
PostgreSQL
gRPC
Kafka
AWS
About Me
I'm a backend and distributed systems engineer studying Computer and Data Science at Rutgers. Most of my work lives in the server layer: high-performance infrastructure, services that coordinate across process boundaries over gRPC and Protobuf, and the data pipelines feeding them. I write mostly Go and Python, usually with Gin or FastAPI in front, and I spend a lot of time in SQL - Postgres, MySQL, MongoDB, and pgvector when the problem is semantic search - deployed across AWS and GCP.
What I actually care about is performance. Where the latency is hiding, which query is doing ten times the work it needs to, how a system degrades instead of collapsing when load shows up unannounced. There's a real gap between code that works and code that holds up under production conditions, and closing that gap - through profiling, better API design, and data models that stay correct when things go wrong - is the part of the job I like most.
Outside of school and code, I bowl, box, play basketball, volleyball, and pickleball, and pick up the guitar when I want to slow down. I'm also someone who needs regular walks with music in my ears to think clearly, and I make time for my friends whenever I can. It keeps me balanced against the hours I spend heads down in a terminal.
Tools I reach for day to day
Go
Python
SQL
PostgreSQL
Redis
Kafka
gRPC
FastAPI
Docker
Kubernetes
Nginx
AWS
Experience
A timeline of my journey and the problems I've worked on.
Myntlo
Founder
At Myntlo, we believe every conversation should create clarity, momentum, and action. Our mission is to help people and teams turn meetings into meaningful progress by bringing important ideas, decisions, and next steps back into focus. We build with simplicity, speed, and trust in mind, helping teams stay aligned and move forward with confidence.
Arka
Software Engineer Intern
I own and execute backend infrastructure tickets supporting Arka's core platform, from investigating system failures to architecting fixes and validating outcomes in production. My work includes building Postgres-backed job reliability systems, such as a stale-job reaper with automated requeue logic, and hardening webhook delivery through retry-with-backoff and persistent delivery logging, focusing on correctness and reliability across distributed backend services.
Deepiri
Software Engineer Intern
I contributed to strengthening the performance, scalability, and reliability of backend systems within a distributed architecture. I worked on improving how services handled high traffic, failures, and real-world production load while increasing system observability, focusing on resilient infrastructure that could scale efficiently and operate reliably under pressure.
Outamation
AI/ML Engineer Extern
I built an end-to-end document intelligence system for large, unstructured mortgage PDFs. I focused on layout-aware OCR extraction and Retrieval-Augmented Generation pipelines that enable semantic search and Q&A over complex documents, then wrapped the final system in an interactive chatbot for real-world mortgage automation use cases.
Projects
A collection of projects I've built, from distributed systems to full stack applications.
Belady
Distributed cache in Go whose eviction policy learns to approximate Belady's optimal algorithm: a LightGBM model trained on sampled access traces predicts which object is reused furthest in the future, and the registry streams new models to live nodes without a restart.
- object hit ratio on a three-node cluster, vs 0.8636 sampled LRU
- 0.8702
- of the remaining gap to optimal closed
- 9.4%
- model scoring per eviction
- ~0.5 µs
Conduit
Distributed job queue and lightweight ELT runtime in Go, built so background work survives crashes and restarts: Gin handles intake, Kafka carries jobs, Postgres tracks job state with leases, Redis Redlock provides mutual exclusion, and Prometheus exports metrics.
LinkNest
Multi-tenant link-in-bio service in Go, built so the hard part lives in the data layer: idempotent click ingestion, an append-only event log, and batched hourly and daily analytics rollups, with the whole write path load-tested using k6.

Pulsegrid
Streaming ingest pipeline in Python, built so one malformed record cannot stall the stream: Pydantic validates every Redpanda message, Postgres writes are batched, failures land in a dead-letter table, and an Airflow DAG rebuilds the SQL analytics layer.
NoteTube
Turns YouTube videos and audio into searchable notes, built to keep slow work off the request path: Celery workers absorb the 30-150 second transcription and embedding jobs, a content service talks over gRPC, and search runs on pgvector.

RU My Valentine
Campus matching platform in Python, built for the traffic spike of a campus event: FastAPI handles authentication and compatibility scoring, Postgres with pgvector powers scalable search, and SQLAlchemy and Pydantic keep the data layer typed. Serving 2000+ users.

- More on GitHubExperiments, coursework, and everything still in progress.github.com/JustinK33
Get in Touch
Have a role, a project, or a systems problem you want a second pair of eyes on? LinkedIn is the fastest way to reach me, and email works just as well.