Your engineering partner for production AI.
MedhAI is an AI engineering firm for banks, lenders and payments businesses.
We work with your team on your own systems: we design the workflow, build the agent and its integrations, test it against the failures it will meet, and stay until it is working in production. We take on one or two engagements at a time, and Atul works on each one directly.
Atul Shridhar.
MedhAI is led by Atul Shridhar. For twenty years, Atul has built fintech, payments, data and AI systems in regulated markets, where mistakes cost money and regulators see them.
Payments and banking. At Paytm, he was founding architect of the layer connecting the Paytm app to its payments bank's core banking, card processing and India's payment networks, built for RBI, UIDAI and PCI DSS requirements. He also led merchant payments for Paytm Canada.
Data and analytics. He built Paytm's company-wide data platform: one trusted source for payments, merchants, commerce and banking, self-serve for hundreds of analysts. Earlier, at Exponential, he built the analytics behind an audience product agencies used for media mix and attribution, and the data pipelines behind conversion models and fraud detection.
AI in production. As VP Engineering at Paytm Labs in Toronto, he led the ML-driven growth and personalization platform for Paytm's 50M monthly transacting users and for PayPay in Japan. Most recently, through MedhAI, he led AI platforms and agent adoption at a large payments company: an in-house AI inference platform, voice and chat agents for customer service and retention, and coding agents used by more than a thousand engineers, with security reviews and human approval built in.
He has also led engineering at Atlassian and Hike. He is based in the Toronto area.
AI agents fail in the same places fast-growing systems always did: at the integrations, the approvals and the exceptions. Most teams are meeting that for the first time. Atul has met it before, in payments, where those mistakes cost money.
Start with the work. Build what it needs.
We start with the procedure your team needs to improve, the systems involved and what the deployment must achieve.
From there, we build the workflow and the supporting integrations, infrastructure and controls. We test exceptions, make unresolved issues visible and agree on the rollout and handover with your team.
We work inside your team, alongside the people who run the workflow, so technical decisions match how the work is actually done.
AI in production at a large payments company.
Customer-facing agents, the platforms behind them and the business systems they work with.
Customer-facing agents
Atul led the program that put voice and chat agents into production across inbound and outbound customer service, sales, retention and merchant support, at a scale of tens of thousands of conversations a day.
Business systems
Agents connected to payments, merchant, lending and case-management systems, so they could look up records, take permitted actions and hand work to people.
Model gateway and cost
Atul led the replacement of a contracted model gateway with an in-house one. Ten use cases from eight teams moved onto it in the first month, with cost-based model routing and one view of spend.
Engineering platforms
An agentic engineering platform used across the engineering organisation.
The client's teams built and operated these systems with Atul leading the program. The client is not named on this site; we can discuss the work in detail in conversation.
Who does what.
We bring the engineering. Your team brings the knowledge of the workflow and access to the systems involved.
Together, we agree what the workflow must achieve, build and test it, and plan the rollout. Your team makes the decisions and gets the operating instructions and known limitations it needs after handover.