Harsh Golani

I build AI systems and data tools that move business metrics.

NYU Computer Engineering, May 2026. Built an e-commerce brand from scratch in my final year of undergrad, scaled it to 20,000 customers and $150K in revenue, then moved to US for my Masters to go deeper on the engineering side.

See My Work ↓ Get in Touch →
Scroll
RAG PipelinesLLM Agents PythonFastAPI OpenAI APIAnthropic API PostgreSQLReact Node.jsA/B Testing Vector DatabasesLLM Evaluation Full Stack EngineeringData Analytics RAG PipelinesLLM Agents PythonFastAPI OpenAI APIAnthropic API PostgreSQLReact Node.jsA/B Testing Vector DatabasesLLM Evaluation Full Stack EngineeringData Analytics
Selected Work

Projects built with purpose
and deployed with care

01 / 07AI / ML

Docquery

RAG-powered PDF Q&A — upload a document, ask questions, get answers with source citations.

Built to understand RAG end to end — embeddings, vector search, retrieval, and generation.

Voyage AI embeddings · ChromaDB vector store · Claude Haiku · FastAPI + React
PythonFastAPIReact ChromaDBVoyage AIAnthropic
02 / 07AI / ML · SDE

Dispatch

AI agent that researches any company — searches the web, classifies results, and synthesizes a structured brief.

Built to understand agentic AI end to end — planning, tool use, cost-aware model routing.

Haiku for classification · Sonnet for synthesis · 60% token cost reduction via model routing
PythonFastAPIAnthropic TavilyReactRender
03 / 07SDE

Traqr

Shortens links and tracks every click — where it came from, when, who.

Built because I couldn't trust third-party analytics on paid campaigns.

Handles 10K+ requests/min via Redis caching · Sub-100ms redirect latency
Node.jsRedisPostgreSQL ReactRailway
04 / 07SDE · PM

Applyd

Track every job application, outreach contact, and follow-up in one place — with a daily home screen that tells you what needs attention.

Built because spreadsheets fall apart after 50 applications and nothing tracks both the application and the conversation around it.

Real-time Kanban updates via WebSockets · JWT auth · contact snooze and follow-up tracking
Node.jsPostgreSQLWebSockets ReactRailway
05 / 07Data Analyst

Storemetrics

Cohort retention, revenue trends, and churn signals for e-commerce.

Built from actually running one — the metrics I wish I had at the time.

Analyzes 100K+ orders · D1/D7/D30 retention with A/B significance testing
PythonPandasPlotly Dash SQLStreamlit
06 / 07Data Science

Columbia Conversion

ML-powered conversion analytics for Columbia Sportswear Turkey's e-commerce site.

Built to translate DataRobot model outputs into actionable business recommendations.

12,330 sessions · Keras Neural Network AUC 0.9306 · SHAP feature importance
PythonPandasPlotly StreamlitDataRobot
07 / 07PM

Applyd PRD

Full product spec for a job search tracker built for tech job seekers running aggressive, high-volume searches.

Written because building without a spec first is how products get rebuilt twice.

3 user personas · 20 V1 features · V2 roadmap · success metrics
Product StrategyUser Research MetricsRoadmap
Background

Engineer who has
shipped real things

I started my first company in my final year of undergrad. A custom phone case brand that grew from an Instagram page to 20,000 customers and $150K in revenue. Running it taught me how businesses actually work: growth, retention, unit economics, and when to stop building something.

I came to NYU for one reason. To go deeper on the engineering side and get exposure I couldn't get staying comfortable. Now I build AI systems and data tools.

Most engineers at my level have only built for users. I have been one.

20K
Customers
from zero
$150K
Revenue
generated
MS Computer Engineering
New York University · May 2026
New York, NY
Technical Skills

What I build with

AI Systems

RAG pipelines, LLM agents, evaluation frameworks, cost optimization

OpenAI APIAnthropic API ChromaDBFastAPI DockerLLM Evaluation
Data and Analytics

Cohort analysis, A/B testing, business dashboards, SQL pipelines

PythonPandas PostgreSQLPlotly Dash DataRobotStatistical Analysis
Full Stack Engineering

REST APIs, auth systems, real-time features, cloud deployment

Node.jsReact PostgreSQLWebSockets AWSRailway
Contact

Let's build
something

Open to AI Engineer, SDE, Data Analyst, and FDE roles
Available May 2026