00 / Positioning

Div
Agarwal

AI/ML engineer building intelligent systems across research and product. NYU CS. Full-stack execution.

I work in the gap between a model that performs in a notebook and a system that holds up when real users touch it—retrieval pipelines, ML workflows, agents, and the interfaces around them.

01Input vectorsQ_32 / V_08
02Embedding fieldDIM / 1536
03Retrieval nodesTOP_K / 12
04Grounded outCITE / TRUE
Pipeline activeCoords [32, 114, 09]

01 / Experience

2020 — Now

Apr 2026 — Present
San Francisco

AI Engineer (Contract) · Daylit

Building ML risk prediction, real-time Airflow pipelines, and conversation intelligence for accounts-receivable collections.

Risk models
Mar 2026 — Present
San Francisco

AI Engineer (Contract) · Variant Avatars

Improving agentic retrieval and multi-agent financial research systems across market data, audio, and presentations.

+30% relevance
Feb — Mar 2026
San Francisco

Software Engineer · AdsGency AI

Built natural-language analytics over DuckDB and ranking systems for multimodal advertising data.

NL → SQL
Oct 2024 — Feb 2026
New York

Researcher · NYU Neuroinformatics Lab

Trained neural event-detection models and shipped a browser tool for exploring calcium-imaging traces.

0.92 F1
Feb — Jul 2023
Toronto

Data Analyst · Adeptmind

Scaled data-driven landing pages from 10K to 80K and built performance monitoring across 19 clients.

+20% traffic
Nov 2022 — Feb 2023
New Delhi

Research Engineer · Enord

Developed deep-learning depth estimation for autonomous drone navigation on ROS and NVIDIA systems.

Depth / ROS
Sep 2020 — Feb 2022
Vellore

Head of Autonomous · Team OJAS · Formula Student

Led perception, planning, and control; helped the team become the first Indian qualifiers for Formula Student Hungary.

92% vision

02 / Selected systems

Full archive →
TIME / NEURAL RESPONSE

Neuro Window Explorer

Interactive calcium-trace analysis for neuroscience research, paired with an event detector reaching 0.92 F1.

Next.jsPythonML
CORPUSRETRIEVEANSWER

Expert-Call RAG

A local, source-grounded retrieval system for searching and reasoning over expert-call transcript archives.

LlamaIndexLanceDBGradio
FETCHVISIONCLAUDEINDEX

Social Automation

A serverless pipeline connecting media retrieval, computer vision, Claude captioning, and OpenSearch.

AWS LambdaBedrockPython
INGESTSCOREPLANACTION

Collections Strategy

An AI product that turns Excel and PDF inputs into transparent, multi-step collections strategies.

FastAPIReactOpenAI

03 / Capabilities

Tools follow problems

Intelligence

  • PyTorch
  • Hugging Face
  • RAG + LlamaIndex
  • LangGraph
  • Computer Vision

Systems

  • Python
  • FastAPI
  • SQL + DuckDB
  • AWS
  • Docker

Interfaces

  • TypeScript
  • React + Next.js
  • Data visualization
  • Streamlit
  • Git

04 / Education

New York · Vellore · Delhi

2023 — 2025

New York University

MS · Computer Science

GPA · 3.76 / 4.0

Machine learning, vision, NLP, big data, cloud computing

2019 — 2023

VIT University

BTech · Computer Science

GPA · 8.4 / 10

AI/ML, statistics, finance, business management

2017 — 2019

Delhi Public School, R.K. Puram

Science + Mathematics

GPA · 92.8%

New Delhi, India

05 / Elsewhere

Beyond the terminal

Long walks, orchestral concerts, ambitious cooking, classic fiction, and records best heard from beginning to end.

Currently in rotation
Dire Straits
Pink Floyd
Steven Wilson
Pearl Jam

06 / Contact

San Francisco · Remote

Build something useful.

Have a hard data problem, an intelligent product, or a research idea? Send the brief—or book a focused conversation.

[email protected]

Email or phone is required so I can reply.