CO-FOUNDER / ANISHFYI.COM
Anish Kr Singh.
Backend and applied AI engineer. Co-Founder of Velofy, infrastructure at Chert (Y Combinator P26), previously FableRoom and Fyle. Working from Delhi, India.
A little about me
I build backend systems and tools for AI. Currently at Chert, a Y Combinator company that gives AI agents their own iMessage and FaceTime lines. Before that, FableRoom, where I built Maestro, its operator agent, and Fyle.
My work spans agent context, document parsing and web extraction. Most of it is open source here at Velofy, the lab I co-founded in Delhi. I write about engineering, applied AI and taste on anishfyi.com, and those essays are listed below.
Work
- Currently building at
- ChertInfrastructure, Y Combinator P26
- Co-Founder
- Velofy
- Previously
- FableRoom, Fyle
- Working from
- Delhi, India
Chert ↗
Now · Y Combinator P26Infrastructure Engineer. The Mac fleet that gives AI agents their own iMessage and FaceTime lines: a software audio route that replaced the physical loopback, the fleet monitor and updater, and the command-line interface (CLI) that puts an agent on a line.
FableRoom ↗
February to November 2026Sole backend engineer for a home decor brand and its business-to-business (B2B) arm. I built Maestro, the operator agent the company asks in Slack and on the web, and Querion, a read-only data analyst that is now open source. Backed by Z47 (formerly Matrix Partners India), WaterBridge Ventures and Alteria Capital.
Velofy
Co-Founder · DelhiAn applied AI and security lab. The public work spans agent systems, document extraction, web retrieval and AI security, all documented on this site.
Fyle
PreviouslyMember of Technical Staff on the integrations team, October 2022 to March 2025. I built the connections between Fyle and accounting systems such as Sage Intacct, Sage 300 CRE, QuickBooks and Xero: the backend that moved the data and the screens finance admins used to map and export it. That means journal entries (JEs), expense reports (ERs) and charts of accounts (COAs) every day. Fyle was backed by Tiger Global, Steadview Capital and Freshworks, and was acquired by Sage in 2025.
Selected work
Open source projects from the lab, with their documentation on this site.
curl_reap
Python scraping with a browser-exact fingerprint.
Pythonterbium
Parse documents into rows, using a model if unsure.
Pythontrove
Repository index and session memory for coding agents.
Rustquerion
A read-only analyst for questions about your data.
Pythonglep
Indexed grep and glob for coding agents.
Rusttroy
A Chromium browser your agent joins, already signed in.
JavaScriptWriting
20 essays on engineering, taste and startups, published on anishfyi.com.
Why I hand-wrote a JavaScript framework in 2026
Why summit.js puts Alpine-style markup on fine-grained signals and a hand-written expression interpreter that runs under a strict Content Security Policy.
Why your scraper gets a 403 and your browser does not
The block usually happens in the TLS handshake, before any header is read. On handshake fingerprints, self-healing selectors and a hand-written crawl engine.
Parse first, escalate only what you must
Most document pipelines send every page to a model. On rebuilding tables from word geometry, scoring confidence from 0 to 1, and escalating only hard pages.
Building Querion: a Read-Only, Natural-Language Data Analyst
How I let a language model query production data without ever letting it write: layered read-only checks, a sandboxed directive loop and an API allowlist.
Trove: Giving Claude Code a Persistent Memory
A file-based memory index that reloads into context every session, so Claude Code stops starting from zero and each session begins further ahead than the last.
Designing Clean REST APIs with Django REST Framework
How I structure serializers, views, and routers in Django REST Framework so API code stays predictable as a project grows.
Modeling an Order Workflow as a State Machine
Turning a tangle of status flags into an explicit, testable state machine that guards every order transition.
Verifying Inbound Webhooks with HMAC Signatures
Why every inbound webhook needs a signature check, and how to verify HMAC payloads safely in Django.
Killing N+1 Queries with select_related and prefetch_related
Finding and fixing the silent N+1 queries that quietly add hundreds of milliseconds to an API response.
Caching High-Traffic Endpoints with Redis
A practical caching layer for read-heavy endpoints, covering cache keys, invalidation, and stampede protection.
Background Jobs with Celery and RabbitMQ
Moving slow work off the request path with Celery and RabbitMQ, with retries and idempotency that survive failure.
Token Authentication for a Django API
Setting up token authentication for a Django API, from issuing tokens to rotating and revoking them.
Bulk Update and Export Endpoints at Scale
Designing endpoints that update and export thousands of rows without timing out or locking the database.
Decoupling Side Effects with Django Signals
When Django signals keep code clean, when they hide bugs, and how to use them without regret.
Zero-Downtime Database Migrations in Django
Shipping schema changes to a live Postgres database without dropped requests or downtime windows.
Full-Text Search with Elasticsearch and Django
Adding fast full-text search to a Django app by syncing models into Elasticsearch and querying them well.
Generating PDF Documents in Django with ReportLab
Building clean, repeatable PDF documents like invoices and purchase orders directly from Django with ReportLab.
Testing a Django API with pytest
A pragmatic pytest setup for Django: fixtures, factories, database tests, and coverage that actually helps.
Production Observability with Sentry, Prometheus, and Grafana
Wiring Sentry, Prometheus, and Grafana into a Django service so production tells you when it hurts.
Dockerizing a Django App for Dev and Production
A Docker setup for Django that stays fast in local development and lean in production.