Senior Backend / Full-Stack Engineer — Distributed Systems & AI

Worked remotely on an AI-run local-services marketplace for the German market. Owned the outbound AI voice agent end to end, including a Python SIP relay through LiveKit into Gemini Live sessions, codec handling at the SIP edge, and fallback support for a second voice provider.

Built code-level validation gates so names, email addresses, and preferences could only be repeated when present verbatim in provider responses, preventing invented contact data. Developed the acquisition pipeline and audit trail in NestJS and PostgreSQL, with call outcomes derived from journaled activity.

Delivered GDPR and billing correctness improvements and shipped across the monorepo, including NestJS services and the Next.js/React product. Used gated GitLab CI deployments from a single host to separate beta and production stacks.

Joined as a software lead and was promoted to team lead for sendout.ai, a LinkedIn automation platform operated by Prospect Engine LLC. Led a team of 13 engineers, overseeing delivery, code review, and engineering standards.

Built and owned the browser-automation fleet powering the product, operating approximately 5,000 concurrent browser sessions on Kubernetes with scheduled campaign actions. Developed the NestJS and Next.js microservice architecture using Kafka, PostgreSQL, and Redis.

Owned platform resilience, scaling design, and the Kubernetes and Jenkins delivery path. Also delivered client projects across Prospect Engine LLC's wider portfolio.

Built backend services for a consumer digital fitness application using Python, Django, Flask, and PostgreSQL.

Containerized services with Docker, completed backend scaling work to support user growth, and developed engagement-analytics modules used by the product team.

For Amar11, a fantasy sports platform, designed and led Java and Vert.x microservices supporting real-time gameplay and leaderboards on PostgreSQL. Mentored junior engineers on the team.

For Lojens.com, an e-learning platform, designed a custom microservices framework using Java, Vert.x, Django, ZooKeeper, and Apache Ignite. Built frontend applications with Vue and React, as well as a Flutter companion application.

Ran freelance software projects for clients from Dhaka, Bangladesh, including continued work on the BCB Livescore scoring platform originally developed at Pavilion360.

Built an automated SEO and Amazon-affiliate business designed to operate end to end without manual steps. The product progressed beyond prototype stage but was not launched because funding and time ran out.

Built an event-sourced, multi-tenant platform using Scala, Akka, Play Framework, and Kafka. Developed the real-time scoring system used for domestic and international cricket matches in Bangladesh.

The scoring platform supported approximately 25,000 to 30,000 concurrent users during live matches and integrated multiple third-party providers. Worked across backend and frontend systems using Node.js, ReactJS, PostgreSQL, Java, Scala, and MySQL, including the Pavilion official website.