Aftab KhanSenior Software DeveloperDelhi, India
I am a senior software developer in Delhi, building since 2022. I take products from the first sketch to production: web, mobile, the systems behind them and the AI inside them. I lead a team of five and work with AI coding agents every day.
A chat assistant that streams its answer as it is written and works through a question in steps: it searches the library of laws and contracts, the web and the user's own documents, and asks for approval before any action that changes something.
Lawyers ask in plain language and get an answer with sources they can open and read. Other AI tools can use the same search through a public API.
Drafting that turns a short brief into a structured contract, written section by section with reference contracts beside it. Translation keeps the structure of the original, down to the clause numbering, so the two versions can be checked against each other.
Lawyers start from a full draft or a full translation and spend their time reviewing it. A job that fails halfway picks up where it stopped.
The assistant inside Word and Outlook. In Word it reads the contract, proposes edits and applies each one as a tracked change the lawyer can accept, undo or revise. It also reviews the other side's changes, answers comments, compares versions and checks a contract against the firm's playbook.
A lawyer reviews a contract without leaving Word, and every AI edit is a normal tracked change. In Outlook the same assistant reads the message and its attachments and prepares the reply.
One billing engine that holds all four, for self-serve customers and for enterprise contracts. Every AI action reserves its credits first and settles them when it finishes, and every change goes into a usage ledger.
One source of truth for what each customer pays and what each customer can use, and a balance that stays right under simultaneous use.
Search over the watchlists with name matching that allows for those differences and ranks results by how close they are. Bulk screening takes a whole list of names, and tracked parties are checked again as the lists change.
A reviewer sees a short ranked list of likely matches, not a raw search dump, for a single name or a full client list. The same screening is open to other systems through a public API.
A business management system for the whole group: leads, quotes and sales orders, purchasing and warehouses, delivery, HR and payroll, accounting, and cost estimates for electrical panels. It also has portals for suppliers and job applicants.
An order moves from quote to delivery in one place, and stock, payroll and the accounts read the same records.
A native app for iPhone and Android. Employees punch in with their location, apply for leave and overtime, and read their payslips and documents. Managers approve requests from a notification, and anyone can ask the HR assistant a question about company policy.
Requests and approvals happen on the phone, on the day. Fixes reach installed apps as over-the-air updates.
In brief
A chatbot a firm adds to its website. It learns from the firm's pages and uploaded files, keeps each firm's content apart from every other firm's, and answers visitors from that content only.
The platform works out which obligations apply to a company, assigns owners and reminders, and checks uploaded evidence against each obligation before it is marked done.
I run the platform on AWS. The API ships as tagged container images behind a load balancer and a firewall. Document jobs go through queues to workers that scale with the backlog, from zero to ten servers. Email and notifications run on Lambda, defined in code.
A chat assistant inside IBMS that answers employees' questions from the company's own HR documents, on the web and in the mobile app, with the conversation saved.
Internal sites and dashboards that the company's own teams used for their daily work.
An edtech product for engineering college students, and the largest project of my time at J Tech Appz.
Developer since 2022
January 2026 to now
An AI platform for lawyers and compliance teams. I lead a team of five and designed and built its core systems, including the assistant inside Word and Outlook.
November 2023 to now
Modderstack
Contract work on IBMS, a business management system for a group of electrical supply companies: web, API and a mobile app.
April 2022 to December 2025
J Tech Appz
Internal sites and dashboards, and Synapse, an edtech product for engineering college students.
Education
Kalinga University, Raipur