● Applied research · Engineer · San Francisco

Gaurav
Narasimhan.

Engineer and builder, first and always. I spend most of my time writing code, right now across two startups in parallel.

Hands-on, every day. I just love building things, and most of my time goes to code: two startup ideas in parallel, each one shipped by me.

Gaurav Narasimhan
BUILDING AGENTS SINCE2022
2022building agents, three generations of them
3products in parallel, each shipped by me
9public repos: production code and research
150largest organization led, IC to SVP
01 · Open source

Real production code, in the open.

One Roof is developed in public. The next six are small pieces carved out of the products: the parts worth reading, and none of the parts that are the product. Then the research.

FLAGSHIP · OPEN SOURCE · MIT Swift

One Roof

Many coding agents. One place to keep up. You stay in charge.

One place for chorus coding: several sessions each of Claude Code, Codex, Gemini, Kimi and Antigravity, all working together. One Inbox for everything they ask, a History of what they did while you were away, loops that run overnight but never merge on their own, and sessions that keep running when the app is closed. Version 2.0.3, signed and notarized.

Swift 6SwiftUIlibghosttyZigmacOS 14+Apple silicon
roofdSwift · Zig · C
AGENT RUNTIME
A session manager whose sessions survive the app, an upgrade, even the daemon crashing.

One holder process per session owns the PTY; a headless Ghostty terminal in Zig keeps the screen, so reattaching redraws it exactly. The daemon keeps no session state at all.

0session state in the daemon: a restart finds every running session by its socket
citation-verifierPython
LLM TRUST
Catch an LLM's made-up or misread values before they reach a system of record.

Grades every extracted field on a fidelity ladder, checking the value against the document's own text, not the model's quote. Dates, money and durations compared by meaning.

100%of injected citation drifts caught, 0 false alarms on unchanged text
change-matcherPython · ONNX
DOCUMENT CHANGE · ML
Is this a real change, or the same clause reworded?

Aligns two versions of a contract or regulation: BM25 candidates, a fine-tuned MiniLM cross-encoder (23 MB, int8, CPU), a value-and-duty guard and a rule-based fallback.

0.47 → 0.88F1 over a hash diff; cosmetic edits flagged drop from 100% to 4.9%
docket-diffPython
DOCUMENT CHANGE
Tell a renumbered regulation from a changed one, and name the change.

Diffs section by section, so a move from §4.7 to §5.1 is one renumber, not a delete and an add. Five rule-based frames name it: deadline deferred, cadence changed, penalty threshold moved…

5regulatory-change frames, standard library only
llm-recall-probesPython
AGENT EVALS
Does the agent actually remember, or does it just sound like it does?

Scores long-conversation recall: refusals before values, a real figure borrowed from the wrong source, fabrication, and giving in to a user's wrong number.

25 / 25self-test cases, each scorer shown failing on a known-bad answer first
fail-open-scorerTypeScript
ML SERVING
A client for an ML scoring service that can never block the request it serves.

Hard timeout even when fetch ignores the abort, a zod contract, a circuit breaker that half-opens for one trial, and checks that reject a whole bad response.

80 msworst case, whatever the model service does
Byte for byteEach copy is production code, not a rewrite for show.
AllowlistedA script names every published file. Nothing else can leak.
Leak-gatedKeys, hosts, tenants and paths are checked on every rebuild.
VerifiedTests and README examples run from a fresh clone.
Research, in the open

Back to school, on purpose: frameworks change every quarter; the fundamentals compound.

02 · The arc

Three generations of agents. Each one taught me what the next had to be.

Every step was a deliberate bet: back to school mid-career for the fundamentals, a leading enterprise software company for scale, an SEO startup for speed, then my own products.

1
Gen 1 · 2022

Research agents

UC BERKELEY

Back to school mid-career to master the fundamentals. Clinical NLP, embeddings, and the first agents, built to understand rather than to ship.

2
Gen 2 · Enterprise

Handcrafted agents

LEADING ENTERPRISE SOFTWARE CO.

Agents that diagnosed incidents across the SaaS fleet, then wrote and ran their own remediation code. Triage false positives down 30–50%.

3
Gen 3 · Growth

SDK-native agents

SEO STARTUP

A fleet of marketing agents and the platform beneath them. A causal recommender from scratch in about ten days. An agent lab on day three.

→
Now

My own labs

GABBLEGRID + ONE IN STEALTH

Each one on its own platform, each a different kind of agent. What I learn there comes with me.

03 · Now building · The labs

Two products, two kinds of agent, one shared backbone.

Eighteen months of nights and weekends, now a few dedicated sprints to take them live. Consumer to enterprise, each on its own stack, evals as release gates.

LAB 01 · CONSUMERJUST ANNOUNCED

GabbleGrid

A coach for engineers who learned to build before AI.

A daily plan sized to the time you have, four steps from scan to launch, and a recommender that learns what moves you forward. The agent here is a coach: it suggests, explains why, and gets more say only as it earns it.

00 Plan01 Scan02 Study03 Build04 Launch
gabblegrid.ai ↗
THE BACKBONE · SHARED BY EVERY LAB

Zibajoon

One admin and instrumentation portal for all of them: one place to see what's running, what changed and what it cost, across two products on two stacks.

zibajoon.com ↗

LAB 02 · ENTERPRISE · ENERGY · IN STEALTH

An operator for regulated industries.

Starting with energy, where every action must survive an audit. Autonomy earned one approved action at a time.

04 · How I take risk

Three rules for agents. Same three for bets.

01

Autonomy is earned.

Agents get more rope only as they prove they can be trusted, and so do bets.

02

Measure honestly.

A before/after delta isn’t proof; a counterfactual is.

03

Bet on fundamentals.

Frameworks change every quarter; the fundamentals compound.

05 · Recent work

Applied AI, from the reference implementation to the org chart.

Led applied AI for agents

An SEO startup

A fleet of marketing agents and the platform beneath them. Built a causal recommender from scratch in about ten days, and started an agent lab on day three.

~10 dayscausal recommender, from scratch
Day 3agent lab started
Fleetmarketing agents + platform
Led applied AI · SaaS operations

A leading enterprise software company

Owned agents that diagnosed and remediated incidents across the fleet, writing and running their own remediation code.

30–50%fewer triage false positives
Agentsdiagnose + remediate
Fleet-wideSaaS operations
ZERO TO ONE

Assemble the team, set the bar, ship, scale.

RECRUITING

PhD researchers and engineers from Meta, Amazon, Netflix and TikTok.

IC TO EXECUTIVE

From the reference implementation to organizations of 150.

06 · Range

Core engineer. Not only an engineer.

The best systems I've built borrowed from somewhere else: a studio, a shop floor, a payroll. Range isn't a detour from depth. It's where the taste comes from.

{}

Engineer & builder

The core, and where most of my time goes. Agents, ML, data and infra, end to end, building two products at once.

HANDS ON, EVERY DAY
▲

Executive

From IC to SVP, organizations of up to 150, and the hiring bar that comes with them.

SET THE BAR, THEN HOLD IT
$

Small business owner

Customers, cash flow and payroll. The fastest teacher of what “it works” really means.

EVERY SYSTEM HAS A P&L
✶

Artist

Founder of The Stable, a community of artists in San Francisco. Making things, and making room for others to.

TASTE IS A SKILL
07 · Stack

What's on the bench.

What I reach for when I sit down to build.

AGENTS
Anthropic SDKOpenAI Agents SDKLangGraphMCPMem0RAG
ML
LoRADPOGRPOPyTorchvLLMCausal inference
DATA
Postgres / pgvectorQdrantRedisKafkaSpark
INFRA
AWSKubernetesTerraformNext.jsFastAPIDjango