I go looking for the problem underneath the messy data. My clearest example: a refrigeration analysis report built on four sources that didn't agree with each other — MFMs, CTS, temperature sensors, and manual entries — which I reconciled, mapped against the actual refrigeration process, and rebuilt around a pro-rata allocation model. Result: 80–90% more accurate.
I started in Electrical & Electronics Engineering — a B.Tech from KG Reddy College of Engineering & Technology, after a diploma from GPT Kotagiri. That grounding in how power, protocols, and machines actually behave is still how I think about product: I want to understand the panel, the PLC, and the meter before I design the dashboard sitting on top of them.
Since November 2022 I've worked across the full IIoT product stack at Nitro Networks and now Patternlab.ai — starting as a data & technology researcher running factory site surveys, moving into product definition for cloud energy monitoring, and now owning end-to-end delivery of OEE tracking, Energy Management, production scheduling, and AI-driven notifications in a startup ecosystem serving manufacturing clients.
My day-to-day sits at the intersection of PRDs and Python — I write the specs, but I also build the prototypes in Python, React, and MongoDB, which keeps the gap between what engineering can ship and what the factory floor actually needs pretty small.
When sources contradict each other — a meter, a sensor, a manual log — the fix isn't picking one and ignoring the rest. It's finding out why they disagree first.
A dashboard built on a guessed process flow will always be wrong in the same way the guess was. I map how the system actually works first.
Equal or flat allocation is the easy answer and usually the wrong one. Pro-rata based on real usage is more work to build, and it's the one stakeholders can trust.
The refrigeration analysis report depended on four data sources that didn't reconcile with each other: MFMs (multi-function meters) for power draw, CTS (current transformer sensors) for current readings, temperature sensors for cold-storage conditions, and manual entries filling the gaps operators logged by hand. Because these sources disagreed, refrigeration load and cost couldn't be attributed with any confidence across shared equipment and zones — the report existed, but nobody could fully trust the numbers in it.
Before touching the allocation logic, I analyzed the process flow of refrigeration itself — how load moves through the system stage by stage — to understand where each data source should logically sit and where readings ought to match up. That mapping is what exposed where manual entries were quietly compensating for gaps or contradicting what the sensors were reporting, rather than just adding noise on top of otherwise-clean data.
With the process flow mapped, I reconciled the MFM, CTS, and temperature sensor readings against it, using the mapped flow as the reference for which readings should agree and which manual entries could be trusted or corrected. On top of that reconciled data, I designed and implemented a pro-rata allocation model — splitting shared refrigeration energy and cost across the relevant lines and zones based on actual usage share, rather than an equal or flat split that would have misattributed cost to zones that weren't driving it.
This wasn't really a refrigeration problem — it was a data-trust problem. Any report built on multiple sources is only as strong as the reconciliation underneath it. Map the real process first, make the sources agree, then allocate by what's actually happening — not by what's easiest to split evenly.
Real-time Availability / Performance / Quality monitoring, cutting manual reporting 80% for 40+ stakeholders. ₹12L ROI realized via reduced downtime.
Centralized EMS for 120+ connected assets — 99% of manual reporting automated, with real-time cost visibility.
IIoT solutions deployed across 22 manufacturing plants, accelerating customer onboarding.
Customized alerts for machine downtime and predictive maintenance anomalies, enabling same-day interventions.
Discovery, PRD, Agile execution, UAT, GTM, and KPI tracking across two complete product cycles.