Rasel Mridha at Exprovia: Technical Project Manager Experience

My remit is delivery across teams rather than a single project. I hold the scope, the dates, and the tradeoff between them, and I am the person who names which of the three moves when something slips.

The week is built around sprint planning and the conversations that feed it. I turn business goals into sprint scope and milestone dates, size the work with the engineers who will do it, and keep acceptance criteria specific enough that a ticket cannot be reinterpreted at review. Client and stakeholder management runs alongside that, conducted directly in English: status that is honest about a risk while it is still a risk, and scope conversations held early enough that there are still options on the table.

Hiring and mentoring are mine as well. I run the hiring pipeline, the onboarding path a new engineer follows, and technical mentoring on architecture and code review for the developers on my teams.

I own production incidents. When a service goes down I run the response — owners assigned, a communication cadence set with the client, and the change that stops the repeat written up afterward rather than promised.

The rest is commercial: budget allocation across teams, effort and cost estimates, and the technical proposals that turn a prospect into a signed scope.

Every engagement starts with an onboarding conversation I run directly, in English — requirements clarification, scope agreement, and the risk conversation, from the first call to a signed scope. I have run more than 50 of these at Exprovia.

The same discipline carries into delivery. 100% on-time milestone delivery on the programs I led traces back to the onboarding call rather than to anything downstream: a risk named while the scope is still being agreed is one that gets designed around, and a scope change gets priced in front of the client instead of absorbed quietly partway through a sprint.

11 delivery incidents are documented as case studies. Three of them, and the option each one ruled out first:

An ECS container deployment failed silently in production and took a critical microservice with it. CloudWatch logs pointed at a misconfigured task definition and missing environment variables. The engineering instinct is to fix forward and redeploy; I ruled that out, because a forward fix is an open-ended diagnosis window while a customer-facing service is down. I rolled back to the last stable task revision first, restored service, and only then corrected the task definition and the IAM permissions and redeployed behind health check validation. Service was restored in 35 minutes. The deployment checklist and automated health checks that came out of it exist for a specific reason: the failure was silent, and the deploy had reported success.

A production database went unresponsive under peak traffic for a live client. Slow queries and lock contention were the cause. Scaling the instance would have absorbed the load and hidden the queries, so I did not start there. I killed the long-running queries to get the service back, rebuilt indexes against the query plans that were actually running, and put a caching layer in front of the repeat reads. Query performance improved by 80%.

A production VPS began dropping requests under moderate traffic. Profiling showed a PHP-FPM pool misconfiguration and unindexed queries rather than a hardware ceiling, which is why the fix was configuration and not a larger server — the same traffic profile would have failed on upgraded hardware a little later, at a permanently higher monthly cost. I tuned Nginx worker processes and the PHP-FPM pool, added a Redis caching layer, and indexed the slow queries. Average response time went from 4.2s to 0.6s, and the server now carries 3x its previous peak load on the same hardware.

Manual reporting and client updates were taking 12+ hours a week. Three options were on the table: hire coordination headcount, cut reporting frequency, or remove the manual step. Cutting frequency trades client trust for team hours, which is the wrong trade on accounts that escalate when they feel uninformed. Headcount would have added a full-time person to retype data that already existed in ClickUp. I built an agent on the OpenAI API that reads ClickUp project data, drafts the stakeholder summary, and delivers it through Slack. Weekly reporting time fell by 80%, and the team reclaimed 10+ hours a week for delivery work.

The Exprovia engineering team went from 7 to 20 under my hiring and onboarding ownership. A team of 7 coordinates by proximity — everyone hears everything, and process reads as overhead. That stops working somewhere in the teens, and the failure mode is invisible until delivery quality has already slipped. So I built the structure ahead of the headcount rather than after it: a hiring pipeline with a consistent technical bar so screening did not drift as volume rose, an onboarding flow that hands a new engineer their environment, their first scoped piece of work, and a named person to ask, and a mentoring cadence that puts architecture and code review feedback in front of developers from someone who reads their code. Growth at that multiple is a systems problem before it is a recruiting problem.

The on-time figure is a scoping outcome rather than a heroics outcome. Dates hold when the estimate came from the engineers doing the work, when a scope change is priced in front of the client instead of absorbed quietly, and when the risk conversation happens in the sprint the risk appears rather than in the week of the deadline — which starts in the onboarding call, not after it.

Project Manager — Exprovia Rajshahi, Bangladesh. On-site, with remote delivery for clients inside and outside the country. Jan 2026 — Present

Associate Project Manager — Exprovia Rajshahi, Bangladesh. On-site. Aug 2025 — Feb 2026

Full Stack Web Developer — Exprovia Rajshahi, Bangladesh. On-site. May 2025 — Aug 2025

Web Developer, Internship — Exprovia Remote. Mar 2025 — May 2025

Total tenure at Exprovia: 1 yr 5 mos. Four roles, one company, seventeen months from first day as an intern to Project Manager.

Education Diploma in Engineering, Computer Science — Satkhira Government Polytechnic Institute Jan 2021 — Dec 2025. Grade 3.46.

The full-stack engineering background this role rests on — MERN, Next.js, PHP and WordPress, SQL and PostgreSQL, AWS and Linux server administration — was built in the two engineering roles that came before the PM ones, at the same company and on the same teams I now run delivery for.

That is the reason the delivery work runs without a translation layer. The architecture calls I review, the estimates I size, and the incident bridges I take are in stacks I shipped in myself, recently enough that the details have not gone stale.