I build software at the intersection of machine learning, agent infrastructure and blockchain. Since December 2025 that has meant Cruq AI, which I cofounded and where I am CTO.
Before it I was CTO at Sendit Markets, and spent two years at Persistence Labs on liquid staking, money markets and decentralised exchanges. Further back: packet processing at Intel, machine learning research at the Centre for Neuroscience, an MTech in computer science at IISc — and, before any of the engineering, two years selling prepaid telecom across Mumbai and Kolkata, which taught me more about how a product actually reaches people than anything since.
Building
-
Cruq AI
2025 —
A visual canvas for composing, deploying and monitoring production AI agents. Bring your own provider keys; we take no cut of your tokens. Underneath sits a multi-tenant execution layer with sandboxed runtimes and streaming traces, and it self-hosts for teams with data-residency requirements.
The bottleneck was never model capability. It was the distance between an agent that works in a prototype and one you would leave running in production.
- talentreview.ai Applicant tracking
- findprop.ai Real estate
- clawduck.com AI assistant
- blinkslabs.xyz Agents for blockchain
Work
-
CTO, Sendit Markets
2024 – 2025
Decentralised payment infrastructure: real-time pipelines on RabbitMQ and Elasticsearch, Go and Python services, and contracts for payments and reputation scoring.
-
Blockchain Developer → Senior, Persistence Labs
2022 – 2024
Liquid staking, money markets, decentralised exchanges and data indexing.
-
Software Engineer, Intel
2021 – 2022
Poll-mode drivers and packet pipelines in C, profiled and benchmarked with DPDK.
-
ML Research Assistant, Centre for Neuroscience, IISc
2020 – 2021
Master’s thesis: brain age and gender estimation from functional and anatomical MRI, using compact 3D and 2D convolutional networks with transfer learning, down to a mean absolute error of 1.99 years on the RADC-MG cohort.
-
Data Science Intern, ZS
2020
Hierarchical demand forecasting on the M5 retail benchmark — 42,840 nested daily series across items, stores and departments. Built the training and validation pipeline, engineered features, and compared forecasting approaches over a 28-day horizon.
-
Management Trainee → Assistant Manager, Sales, Vodafone Idea
2016 – 2018
Territory manager for prepaid sales in Mumbai and Kolkata, running a distributor network through ₹8 million of business a month with 20 field sales executives.
Projects
-
Covid-19 tracker
2020
Android app for real-time case counts and essential services across India, pushing a notification as the numbers moved.
-
Sudar
2016
Android app bringing efficiency and transparency to the channel distribution system in the Tamil Nadu and Chennai circle, for around 500 users. Built while I was selling for Vodafone Idea.
Education
-
Indian Institute of Science
2019 – 2021
MTech, Computer Science.
-
Indian Institute of Management, Indore
2014 – 2016
MBA, Finance and Marketing.
-
IIEST Shibpur
2010 – 2014
BE, Electronics and Communications Engineering.