0%
Open to LLM / Generative AI Engineer roles

Kushal Kothari

_

I build LLM systems that work in production — RAG pipelines, fine-tuned models, and agentic workflows. MS in Computer Science from NYU · Data Analyst at Optimum · Google Summer of Code alum.

View my work → Get in touch
GitHub LinkedIn Medium Scholar
scroll

01 — About

Engineer first.
AI-obsessed.

From backend systems to LLMs in production.

I'm an AI engineer with a CS foundation from NYU and hands-on experience building LLM-powered systems — RAG pipelines, fine-tuned models, and agentic workflows — using LangChain, the OpenAI API, and Hugging Face Transformers. Comfortable across the full stack, from vector database design (Pinecone, Chroma, Weaviate) to production deployment.

Previously at Google (Summer of Code) working on network systems with the Erlang Ecosystem Foundation. Published researcher in ML and computer vision (IEEE ICICT 2022). Currently at Optimum, building data pipelines on GCP — deepening my cloud and data infrastructure foundation while shipping AI projects on the side.

What I build: systems where LLMs actually work in production — low latency, grounded outputs, minimal hallucination.

Kushal Kothari
New York, NYest. 2023
0

Peer-reviewed publications

0+

Users served by LLM pipelines I built

0

Networks analyzed for WBD reporting

0+

Students mentored in open source

AI / LLM

LangChainRAGOpenAI API HF TransformersLlama 2 fine-tuning Prompt EngineeringMLflow PineconeChromaWeaviate

Data & Cloud

BigQueryGCPSnowflake DatabricksAWS AthenaAWS Glue LambdaKafkaDocker

Languages

PythonSQLJavaScript TypeScriptJavaC++

02 — Experience

Where I've
shipped.

Data AnalystCurrent

Feb 2026 — Present

Optimum · New York, NY · Full-time

  • Engineer BigQuery SQL pipelines processing NRP-weighted viewership data across 16-month tuning windows — audience overlap analysis for NFL, NBA, MLB, and NHL properties.
  • Lead analytical workstreams for the Warner Bros. Discovery client: HBO MAX authentication impact, wholesale subscriber behavior modeling, and cross-platform reporting across 24 O&O networks.
  • Built HBO linear tuning analysis showing authenticated households watched 14–16% more HBO linear content — directly informing WBD's wholesale retention strategy.
  • Supporting Databricks migration of Promo Analysis, NRP, and MAX Reporting pipelines across Snowflake and BigQuery, plus the AWS Athena → GCP BigQuery migration.
BigQuerySQLSnowflakeDatabricksGCP

Data Analyst Intern

Oct 2025 — Feb 2026

Optimum · New York, NY · Internship

  • Led an end-to-end metric validation framework for feedzone logic, quarter-hour segmentation, and broadcast month derivation — ensuring statistical accuracy for C-suite reporting.
  • Built automated data quality checks reducing manual validation overhead across high-volume streaming datasets.
BigQuerySQLData QualityAWS Athena

AI Engineer Intern

Sep 2025 — Nov 2025

SalesCaptain · New York, NY · Internship

  • Fine-tuned Llama 2 on domain-specific sales conversation data using Hugging Face Transformers, improving response relevance for automated customer engagement.
  • Built a RAG pipeline with LangChain and Pinecone to ground LLM outputs in company-specific knowledge bases, reducing hallucinations in customer-facing AI.
  • Engineered prompt-chaining pipelines for automated follow-ups, reminders, and customer insights — serving 500+ users across SMS and webchat.
  • Integrated the Twilio API with LangChain-orchestrated LLM workflows for real-time AI-powered SMS automation.
LangChainPineconeLlama 2TransformersTwilio

Information Technology Engineer

Jan 2024 — May 2025

NYU Wasserman Center for Career Development · New York, NY · On-campus

  • Developed a secure ticketing system in Python with integrated authentication, reducing response times by 50%.
  • Optimized NYU's cloud infrastructure (+30% scalability) and built real-time analytics pipelines for internal performance monitoring.
PythonAWSAuthenticationAnalytics

Product Analyst

Sep 2022 — Aug 2023

Media.net · Mumbai, India · Full-time

  • Led website migration projects and performance analysis on relational databases — 20% fewer launch errors, saving $32,000.
  • Primary technical liaison for product issues, cutting support ticket resolution time by 30%; documentation drove a 15% lift in user engagement.
  • Optimized AWS cloud infrastructure for performance and cost efficiency.
PythonSQLAWSAdtech

Google Summer of Code

Summer 2020

Google · Erlang Ecosystem Foundation · Remote · Open Source

  • Worked on network systems for the Erlang ecosystem under Google/EEF mentorship — open-source contributions around the Yggdrasil encrypted mesh network.
  • Earned a recommendation letter from my Google mentor for dedication and technical contributions to open-source software.
ErlangNetworkingOpen Source

03 — Projects

Selected
work.

001↗

Real-Time Trip Analytics Platform

End-to-end ETL pipeline processing ride data into BigQuery for real-time analytics across revenue, distance, and usage trends. Automated workflows cut manual effort at scale.

BigQueryETLSQLPython
002↗

Stock Market Streaming Pipeline

Kafka + AWS Glue streaming pipeline processing 100K+ trade records in real time. Analytics-ready datasets in Athena with Python/SQL transformations.

KafkaAWS GlueAthenaPython
003↗

Advanced Malware Detection System

ML model classifying 21,736 malware files into 9 families with XGBoost. Optimized feature extraction from 200GB of data, cutting processing time by 35%.

XGBoostPythonML
004↗

AWS Serverless Microservices

Serverless microservices on AWS Lambda, API Gateway, and DynamoDB. RESTful API design optimized for scalable, cost-efficient cloud architecture.

LambdaAPI GatewayDynamoDB
005↗

Tor vs Yggdrasil Network Analysis

Comparative study of encryption and anonymity in Tor versus the decentralized Yggdrasil Network, proposing Yggdrasil as a future-proof alternative. Published by IEEE (ICICT 2022).

NetworkingPythonIEEE
006↗

Real-Time Object Detection

End-to-end deep learning combining SSD with MobileNet for efficient real-time detection and tracking on low-compute devices. Published in IJRASET (March 2022).

SSDMobileNetComputer Vision

04 — Research & Credentials

Proof of
work.

Publication · IEEE

Tor vs Yggdrasil: A Comparative Study of Two Different Communication Systems

Analyzes the encryption and anonymity features of Tor against the decentralized Yggdrasil Network.

IEEE · April 2022 Publication · IJRASET

Real-Time Object Detection using SSD and MobileNet

Deep learning approach to end-to-end object detection for devices that lack computational power.

IJRASET · March 2022
Certification

AWS Certified Solutions Architect — Associate

Amazon Web Services

Issued Jun 2025 · Valid through Oct 2030
Education

MS, Computer Science — New York University

Machine Learning & AI, Distributed Systems, Cloud Infrastructure & Security. Graduate Life Ambassador.

2023 – 2025 · New York, NY
Education

BTech, Computer Science — University of Mumbai

Operating Systems, DBMS, Software Engineering, AI. Lead of the Competitive Programming Team.

2018 – 2022 · Mumbai, India
Recognition

Google Mentor Recommendation Letter

Awarded by my GSoC mentor for impactful open-source contributions with the Erlang Ecosystem Foundation. Also mentored 150+ students via AnitaB.org.

Google Summer of Code

05 — Writing

Notes &
deep dives.

Engineering

Building a Real-Time Trip Analytics Platform

End-to-end walkthrough: from raw ride data to BigQuery-backed dashboards.

↗ Engineering

Building a Real-Time Stock Market Streaming Pipeline

Kafka + AWS Glue architecture for processing 100K+ trade records with low latency.

↗ Essay

"Even death has a heart." — Markus Zusak

A reflection on The Book Thief — beauty, brutality, and the human condition.

↗ Essay

"For you, a thousand times over." — Khaled Hosseini

On redemption, loyalty, and friendship in The Kite Runner.

↗

06 — Contact

Let's ship
something real.

I'm open to LLM / Generative AI engineer roles where I can ship fast and go deep. If you're putting language models into production, I'd love to talk.

kushalkothari285@gmail.com
LinkedIn GitHub Medium Scholar Résumé
↑