Arjun Aggarwal

Founding engineer at Lightfield, building agentic CRM and customer-context infrastructure in San Francisco.

I’m interested in systems that turn scattered context into useful work: agents, APIs, workflow automation, developer tools, and the infrastructure around customer-facing teams.

contexttoolsworkflow statereviewaction

Selected work

Where I've worked

all work

Lightfield

2025–present

Founding Engineer

Building infrastructure for agentic CRM: public APIs, workflow automation, agent tools, human-in-the-loop review, notifications, and core customer-facing product surfaces. The work is about making customer context actionable for both people and agents.

Amazon Web Services

2024

SDE Intern

Built financial reconciliation infrastructure for high-volume payment events. The work was about reliable systems: tracing failures, preserving reporting completeness, and making financial data trustworthy as it moved across services.

Capital One

2023

ML Engineering Intern

Worked on graph ML infrastructure for large-scale card and customer relationship data. I spent a lot of that internship studying how Twitter and Facebook modeled social graphs, especially systems like TAO, and started to see how much product intelligence comes from making relationship data queryable. This is where the context-graph thread started.

Selected writing

What I'm thinking about

all writing

Agent Harnesses, Not Chatbots

forthcoming

Why the next useful AI products will look less like chat windows and more like workflow systems with a model inside.

A short note on why useful agent products feel less like chat windows and more like quietly maintained workflow graphs.

Why Customer Context Is Becoming Infrastructure

forthcoming

A short essay connecting Lightfield, CRMs, forward-deployed engineers, and why the customer system-of-record is up for grabs again.

From the log

Lately

full log

Jun 7 · thought

Agents need workflow state: The best AI products don't feel like chatbots. They feel like someone quietly cleaned up the workflow graph behind the scenes.

Jun 7 · album

Mk.gee, Two Star & The Dream Police: Night-driving music for a city you don't live in anymore. Still growing on me.

Jun 6 · film

Heat: Competence, loneliness, and LA at night. The diner scene still does more with two men talking than most films do with everything.

Jun 6 · restaurant

Trèsind Studio: Technically ridiculous. The pacing and the room mattered as much as the food.

Jun 5 · link

Agent evals and fake precision: Saving this because it explains why agent evals often become fake precision: clean numbers on a benchmark that doesn't resemble the real workflow.