Invited speaker at CloudX 2026 (Santa Clara, Sept 1-3), co-located with API World + AI TechWorld. Speaking on how to forecast hardware capacity for AI infrastructure at billion-user scale.
How a technical program manager actually runs a large program, from the first funding conversation to the hundredth status update.
Mashable profiled my work leading AI infrastructure at billion-user scale: lifting model deployment success from 60% to 99%, cutting scale-up time from three days to two hours, and a GPU efficiency redesign worth an estimated $20M in savings.
Invited speaker at the Kong API + AI Summit 2026 in Los Angeles (Sep 30 – Oct 1). A practitioner session on the operational realities of running large-scale AI inference: deployment coordination, GPU and infrastructure constraints, performance tuning, and reliability under rapid model iteration.
Every technical program manager eventually runs into the same uncomfortable fact: you are accountable for outcomes you cannot command. The engineers do not report to you. The architects do not need your sign-off. The executive who set the deadline will not write a line of the code that meets
Upcoming: I'm speaking at AI Infra Summit 2026 (Santa Clara, Sept 15-17) on what it takes to run AI infrastructure at billion-user scale.
I'm serving as Session Chair at the Sixth IEEE International Conference on Intelligent Technologies (CONIT 2026) in Hubballi, India, on June 20, 2026.
My paper on a replica-centric capacity-planning framework for GPU and Multi-Instance GPU (MIG) based AI inference platforms was accepted at IEEE MetroInd 4.0 & IoT 2026 in Rome, Italy.
I spoke at the Budapest Data + AI Forum 2026 on why ML deployments fail in production and how to engineer reliability at scale. Session abstract, key takeaways, and links.
I recently sat down with Business Standard for their Manager's Mantra series (BSmart). What started as a conversation about automation and managerial roles turned into a wide-ranging discussion: enterprise cloud strategy, the economics of AI infrastructure, how to build data platforms people actually use, modernizing legacy systems,
I presented our paper "Operationalizing Site Reliability in Large-Scale Distributed Systems: Shifting Ownership Left" at IEEE CSNT 2026 (Al-Khobar, Saudi Arabia): predictive models that forecast failures to keep large-scale distributed systems reliable.
OneIndia profiled my work on AI model deployment: cutting deploy times from days to hours, lifting success rates from roughly 60% to nearly 99%, and serving about 4x more requests per GPU.
Highlights
I joined a Zee Business India 360° panel (Feb 2026) on AI hallucinations: why models confidently make things up, how to verify their output, practical prompt techniques, and where AI regulation stands across the EU, US, and India.
System Design
Every database we've met so far stores facts: a user's name, an order's total, a sensor's reading, a word in a document. The newest member of the family stores something stranger and more powerful: meaning itself. Vector databases are the data system
System Design
There's a kind of data that has quietly exploded in volume over the last decade, and you generate it constantly without thinking. Every time a server reports its CPU usage, every time an app records how long a request took, every time a thermostat notes the temperature, every
System Design
Type a few words into a search box and, in a fraction of a second, you get back the most relevant results out of millions or billions of documents, ranked best-first, with your typos forgiven and suggestions appearing as you type. We're so used to this that
System Design
Every database we've covered so far has one job: running an application. They store orders as they're placed, serve profiles as they're viewed, accept writes as they happen. This is the live, operational heartbeat of a product, and it's called OLTP, online
System Design
Every database family we've covered so far treats relationships as a secondary concern. Relational databases reconstruct them with joins at query time. Document databases mostly avoid them by nesting data. Key-value and wide-column stores barely acknowledge them. But for a whole class of important problems, the
System Design
After the elaborate machinery of relational, document, and wide-column databases, the key-value store is almost a relief. It is the simplest possible database, and that simplicity is exactly where its power comes from. There is no schema to design, no query language to learn, no joins to reason
System Design
Some data does not arrive politely. It pours. Think of every tap and swipe in a mobile app, every message in a chat platform, every reading from a fleet of sensors, every location ping from a fleet of vehicles, every line of a log from thousands of servers. This is
System Design
The relational database we covered in the last deep-dive is brilliant at one thing above all: taking your data apart into tidy, non-repeating tables. That is its great strength. But it is also, for some kinds of work, its great inconvenience. A lot of the time, the thing
System Design
If you only ever learn one kind of database deeply, make it this one. The relational database has been the default way to store important data for over forty years, and despite a parade of trendy alternatives, it is still the right answer for the majority of systems being built
System Design
If you have ever looked at a modern system's architecture diagram, you have probably wondered the same thing every newcomer wonders: why are there so many different databases in here? There is a relational database over here, something called a document store over there, a cache, a search
TPM Knowledge
For years, the question “TPM vs PM vs EM” quietly stressed me out. Not in a theoretical way, but in a very real “I have no idea what I actually want to be” way. Not because I was overthinking titles, but because my early career created a lot of mixed