Karthik Gooli

01 Experience

Foxpass · Splashtop ↗ Acquired by Splashtop · 2023

Mar 2021 — Present

Sr. Staff Engineer

San Francisco, CA (Remote)

YC startup providing server and WiFi authentication as a service to SMBs. Key contributor to reaching multiple millions in ARR.
Own the homegrown LDAP cluster serving 20–25k QPS with 150k persistent TCP connections (Twisted). Implemented context-aware search indices — latency reduced several-fold, CPU down 30–40%.
Built clustered LDAP with partitioning — cut per-node memory 40%, DB reads 75%. Modified open-source libraries; serving hundreds of customers. GitHub ↗
Re-architected global RADIUS infrastructure behind a global accelerator with lightweight PoPs worldwide, significantly reducing authentication latency across geographies.
Designed and shipped passwordless WiFi auth for Chromebooks — replaced Google's non-scalable GCCC alternative. Zero manual intervention; opened a major new revenue stream.
Built port-based RADIUS proxy from scratch in async Python; in production at hundreds of customers.
Built RadSec (RADIUS over TLS) proxy handling 10k+ concurrent connections on a single instance. Blog ↗
Implemented eduroam for UDP and TLS. Blog ↗ Built Global Lookup — routes new users to nearest regional stack (US/EU/AU).
Replaced LDAP logging backend with DocumentDB — query results 10× faster, no visible frontend latency.
Open-source: smallstep/pkcs7, PyScep, asn1crypto, FreeRADIUS proxy protocol. Published TTLStore on PyPI.

PythonTwistedLDAPRADIUSGolangPostgresDocumentDBECSDockerDjangoAWS

Peritus.ai ↗ Acquired by Essenvia

Jan 2018 — Feb 2021

Founding Engineer · Tech Lead

Palo Alto, CA

2nd engineer at the company; grew eng team to 5. Patent US10679133B1 ↗
Designed and built a horizontally scalable multi-tenant data ingestion platform from scratch — real-time data from 300+ community sites (Stack Overflow, GitHub, Discourse) using S3/GCS, Athena/BigQuery, and Postgres.
Created libraries to read and normalize data across Athena and GCP, exposing a unified view feeding the ML/IR pipeline.
Modified open-source rate limiter divvy (NodeJS) to accept live config changes; deployed at scale on Kubernetes.
Conceived and shipped a Rules Engine end-to-end — wireframes through full backend — enabling rules over terabytes of machine data.
Productionized SourceGraph as in-house code search across thousands of repositories.
Evaluated and scaled Dgraph (knowledge graph) and Solr; evaluated Kubeflow for model training and deployment.
Introduced MongoDB as a feature store for ML pipeline steps; introduced Redshift for large-volume S3 data.

PythonJavaKubernetesS3AthenaBigQueryPostgresMongoDBRedshift

Fisdom ↗ Acquired for $150M

Jan 2017 — Jan 2018

AVP, Engineering

Bangalore, India

Built backend for the wealth management app. Set up engineering best practices — code reviews, CI (Jenkins, JMeter), alerting, monitoring, Sentry.
Introduced MySQL, cutting data serving latency from 3s → 400ms; rewrote components to leverage it.
Wrote full backend for DIY investment feature; launched top trending funds; introduced Highcharts graphs in mobile apps.
Automated sub-broker commission payments end-to-end with zero manual intervention.
Prototyped AI chatbot (Dialogflow/wit.ai), WhatsApp onboarding (yowsup), and OpenCV image cropping for ops.

PythonJavaScriptMySQLGoogle App Engine

Uber

Jan 2015 — Feb 2016

Software Engineer II

San Francisco, CA

2nd engineer on Kafka platform — grew team from 2 to 20 in 6 months. Promoted within 6 months of joining.
Built and scaled Uber's logging/events infrastructure using Kafka across US and China — 4 clusters of 16 nodes each at 30–40k QPS, inter-DC aware.
Built Golang logging libraries, Kafka 7→8 migration, monitoring/alerting, and timestamp-based Kafka consumer tooling used company-wide.

GolangPythonKafka

Uber

Feb 2016 — Jan 2017

Software Engineer II

Bangalore, India

One of the first engineers in Uber Bangalore — helped grow the office from 0 to 20 engineers.
Designed and shipped the vehicle leasing product ground up; built a distributed framework using the SWIM gossip protocol to scale the service.
Introduced Kafka, Samza, and Spark in Uber Bangalore to solve the India leasing problem at scale.

GolangKafkaSamzaSpark

NVIDIA

May 2014 — Aug 2014

System Software Intern · Android Kernel

Santa Clara, CA

Built a framework in the Android kernel to analyze and correlate kernel events with architectural events; generated visualizations identifying hot-spots and architectural anomalies.

University of Illinois, Urbana-Champaign

Jul 2013 — Dec 2014

Graduate Student Researcher · Systems Research Group

Urbana-Champaign, IL

Built infrastructure to analyze heap growth patterns via VM introspection (VirtualBox + Volatility). Reconstructed raw heap to visualize data structure transformations over time.
Research presented at TCIPG Industrial Workshop 2013 & 2014. GitHub ↗

NetApp India

Jul 2009 — Jul 2013

Software Engineer

Bangalore, India

Kernel and distributed file system development in WAFL (Write Anywhere File Layout).
Authored 4 technical invention papers; received Patent Incentive Award for all four.
Deep work in SAN, SCSI, and FC protocols — root cause analysis and fixes for interoperability issues.
Built i-sense — automated file system damage assessment and recovery, saving significant engineering resources.

CKernelWAFLSANSCSI

02 Education

M.S. Computer Science

University of Illinois, Urbana-Champaign

GPA 3.52013 – 2015

B.E. Computer Science

BMS College of Engineering, Bangalore

79% · GPA 4.02005 – 2009

12th Standard

St. Joseph's College, Bangalore

98.3%2003 – 2005

10th Standard

Cambridge High School, Bangalore

91.7%Till 2003

03 Research & Academic Projects

Heap Analyzer

VM introspection (VirtualBox + Volatility) to analyze heap growth patterns and detect security compromises. GitHub ↗

Helenus

Distributed Key-Value store written in C.

HEVC Intra Frame Prediction

GPU-accelerated implementation using CUDA.

SQLite3 ADD CONSTRAINT

Extended open-source SQLite3 with ADD CONSTRAINT functionality.

Power-constrained Load Balancing

Mitigating heterogeneity across power-constrained nodes in parallel applications.

TTLStore

Python library on PyPI; used in production at Foxpass. PyPI ↗