Hello, I'm Priyansh Sharma/ Backend & Distributed Systems Engineer

I make the complicated parts work together.

I'm a software engineer at Thales. I take complex, open-ended problems from requirements through architecture, implementation, and validation. My work spans distributed systems, backend orchestration, stateful workflows, and AI-powered developer tooling.

I care about understanding the whole problem, making deliberate tradeoffs, and building systems that stay reliable, understandable, and adaptable as they evolve.

BASED IN INDIABACKEND SYSTEMSDISTRIBUTED PROCESSINGAI & DEVELOPER TOOLING

01 / SELECTED WORK

A few problems I've spent time on.

Not a list of everything I've touched. These are the kinds of engineering problems that have shaped how I think and work.

01 / DISTRIBUTED PROCESSINGPROFESSIONAL WORK

Designing the core engine for distributed billing.

Usage-based billing had to apply configured prices and discount tiers per device or across a tenant's usage. Splitting that work across workers meant preserving the pricing rules and knowing when the report was complete.

I independently architected and implemented the core bill-run engine, from mapping through reporting. I used Cassandra-backed coordination to preserve stage progress across worker interruptions and make recovery part of the execution design.

Read the engineering case study
Professional experience with conceptual illustrations; disclosure review remains outstanding.
A SYSTEM, IN STAGES FIG. 01
01MapDistribute work
02ReduceAggregate
03FinalizeBuild output
COORDINATION MATTERS BETWEEN THE BOXES.↗
02 / SYSTEM DESIGNPROFESSIONAL WORK

Redesigning the framework for Fleet interactions.

Fleet decisions and calls were spread across existing Order Scheduler paths. In October 2026, I designed and implemented a centralized Fleet interaction framework, separating side-effect-free planning from execution. I introduced it in stages, with established scheduler and provider-facing behavior as a compatibility constraint.

Read the engineering case study
Professional experience with a conceptual diagram; employer disclosure review remains outstanding.
SEPARATE THE CONCERNS FIG. 02
DECIDEPlannerWhat should happen?
DOExecutorHow does it happen?
ILLUSTRATIVE, NOT AN INTERNAL DESIGN DOCUMENT
03 / BUILDING FOR MYSELFDEVELOPER TOOLING

Helping AI keep up with a real codebase.

AI tools are useful until the context gets messy. I've been experimenting with a persistent engineering knowledge workflow so I don't have to repeatedly explain the same architecture, conventions, and decisions across long-running development sessions.

More details when a public, reproducible version is ready.
CONTEXT OVER TIME FIG. 03
01 Architecture decisions ↗
02 Working conventions ↗
03 Lessons from changes ↗
KEEP WHAT MATTERS. RETRIEVE IT WHEN NEEDED.

A note on these projects: some of my most substantial work has been on private systems. I share the engineering ideas and lessons, not proprietary code, configuration, or customer details.

02 / A LITTLE ABOUT ME

Engineer by profession. Curious by default.

I started out enjoying the straightforward part of programming: solving a problem and getting the right answer. Working on real backend systems taught me to care just as much about everything surrounding that answer.

What happens if the process dies halfway through? Who owns the state? What changes when the workload grows? Can someone else understand the design six months later?

Those questions keep the work interesting for me. Outside my day job, I spend time on algorithms, engineering tools, and learning how good systems are put together.

More about my experience on LinkedIn ↗

03 / THINGS I'VE LEARNED

The details are the work.

A few lessons I keep coming back to. None of them are glamorous. All of them matter.

01

Failure paths deserve first-class attention.

Retries, duplicates, partial results, and recovery aren't edge decorations. They're part of the design.

02

Simple to explain is a useful design constraint.

If a change needs too many exceptions to make sense, I usually need to revisit the boundaries.

03

Measurements need context.

A benchmark is meaningful when you can explain the workload, environment, and what was actually measured.

04 / SAY HELLO

Have an interesting problem?

I enjoy talking about backend engineering, distributed systems, tooling, and the tradeoffs behind building software. Feel free to reach out.