Akhil Singh
I build and ship production AI products. Agentic systems, and the plumbing that keeps them running.
I am a software engineer and operator with 6+ years building revenue-critical systems. Today I build AI products end to end: my own tools for underwriting and document workflows, plus agentic systems shipped for clients.
Credibility
Grounded proof from six years of building systems and products.
Years Building
Experience across platform engineering, product systems, and AI workflows
Systems Supported
Transaction volume supported through production commerce systems
AI Cost Reduction
Model cost efficiency improvement achieved in production work
Products Built
AI products built across underwriting, document workflows, DM sales automation, and a live client project
What I build
Products and client work.
My own AI products alongside work I have shipped for clients, informed by years of backend, platform, and AI systems work.
Sarvada.ai
A product in progress for underwriting teams that need a faster way to review and work through risk.
- The product is aimed at making underwriting analysis more structured and less manual for teams reviewing submissions.
- I am building around practical workflow needs like clearer analysis, reporting, and outputs teams can actually use.
- It is still early, and the emphasis is on finding the right product shape rather than overstating maturity.
Papermill
A live product for turning PDFs and images into cleaner, AI-ready document outputs.
- Built to make document ingestion less brittle for teams working with retrieval and LLM pipelines.
- The focus is on reliable extraction, usable structured output, and fewer manual cleanup steps.
- Like Sarvada.ai, it reflects a shift from operating systems for companies to building products directly.
ReplyJi
An AI agent that answers Instagram and WhatsApp DMs for small online sellers in seconds — in the seller's own voice and grounded in their catalog — so slow replies stop costing sales.
- Built for Indian D2C sellers and small shops that lose orders when DMs sit unanswered — it pulls Instagram and WhatsApp into one inbox and replies in Hinglish, Hindi, or English.
- Answers are grounded in the seller's own catalog and pricing, so the agent handles the repeat questions — price, size, COD — on its own and routes anything it is unsure about to the owner instead of guessing.
- Runs around the clock so late-night, ready-to-buy messages get answered instantly rather than going cold by morning.
FleetDecks
A live product I built for a car rental business — the operating system where the front desk is an AI agent, not a dashboard. Bookings run on WhatsApp and a mobile app while a fleet backend keeps every car accounted for.
- Built for a rental operator to replace spreadsheets and scattered chats with one system covering bookings, vehicles, handovers, and payments.
- Customers book by talking to an AI receptionist on WhatsApp that handles the conversation end to end, while an Android app puts an ordering agent in the team's pocket.
- A fleet backend ties it together so every trip, deposit, and vehicle stays accounted for in real time.
Experience
Where I have worked.
Peve Visions
Software Architect
Bangalore, India
Subscription billing, LLM cost work, and analytics on the Go stack.
- Built the subscription microservice in Go — revocations, unpaid handling, chargebacks. It's what keeps recurring revenue from leaking.
- Cut LLM token usage 5x by refactoring the Go services and the JS inference layer. Answer accuracy went up at the same time.
Teachmint
SDE III
Bangalore, India
Backend for the Teachmint X classroom stack — login, billing, comms.
- Built QR login, device tracking, and subscription flows. Login got 30% faster.
- Shipped in-app purchase validation across Android, iOS, and web. Drove INR 1.5 Cr+ in recurring revenue through it.
Infra Market
Senior Software Engineer
Bangalore, India
Cart, order, and inventory backends for FOFO commerce.
- Built the cart microservice in Go. Powered 19 FOFO stores, $20M+ in transactions, 99.99% uptime. Redis + Postgres underneath.
- Designed the order and inventory systems on Kafka so stock stayed in sync across warehouses and stores in real time.
Yayshop
Founding Engineer
Bangalore, India
Founding engineer — Android + web launches, internal tooling, CRM.
- Shipped the Android and web launches end to end as one of the early engineers.
- Wrote the CRM + analytics layer on BigQuery and Angular Material. Sales query resolution dropped 50%.
What I work on.
Areas of depth
Most of my work sits at the intersection of backend systems, AI workflows, and product execution, with an emphasis on solving practical operational problems.
Operating principle
Start with the workflow, build for adoption, and let technical depth act as execution leverage.