Tonmoy Talukder
I'm Tonmoy Talukder, a backend-focused full-stack engineer and NLP researcher building scalable infrastructure, distributed systems, and low-resource AI solutions.
My engineering work focuses on backend systems that need to be correct, observable, and reliable under real production workloads. I work across API design, database modeling, job queues, rate limiting, authentication, deployment automation, and system monitoring. At WorldTech, I have contributed to production systems including a project serving 30K–50K requests/day.
I care deeply about the fundamentals behind scalable systems: how services communicate, how background jobs are processed safely, how databases evolve without downtime, how failures are detected early, and how engineering teams keep systems maintainable as complexity grows. My current stack includes Go, TypeScript, PostgreSQL, RabbitMQ, Redis, Docker, Prometheus, Grafana, and OpenTelemetry.
In parallel, I research low-resource language NLP and Computer Vision, especially for Bangla. My research is motivated by a simple problem: modern AI systems still perform unevenly across languages, especially for communities with limited high-quality datasets and tooling. I have published work at LREC 2026 and BIM 2023, and I'm interested in building AI systems that are not only accurate in research settings, but also usable in real-world products.
My long-term goal is to build reliable, scalable infrastructure that bridges advanced AI research with real-world products. I'm currently focused on backend and infrastructure engineering roles where system design, correctness, reliability, and practical impact matter deeply.
Experience
2 positions
Software Engineer · Backend / Full Stack
Full-time- Own backend development for production ERP and inventory-management systems, with focus on multi-tenant RBAC, authorization flows, API design, database modeling, and reliable service delivery.
- Designed rate-limiting layer with zero incidents and sub-5ms P99 overhead
- Selected as project lead for BSFIC Store, a government-sector multi-mill inventory management platform for sugar mills.
- Build systems with focus on correctness, maintainability, observability, and operational reliability.
Jr Software Engineer · Frontend / Full Stack
Full-time- Implemented a production document-processing service for a government-sector system, serving 30K–50K requests/day.
- Acted as frontend technical owner for ERP platforms, driving module architecture, reusable components, and production feature delivery.
- Implemented idempotent job processing with at-least-once delivery guarantees
Skills & Tech Stack
Education
Ahsanullah University of Science and Technology
Bachelor of Science in Computer Science & Engineering
Highlights
Low-resource Bangla NLP — Thesis & Publication Track
2022 — 2023- Completed final-year thesis titled “Bangla Key2Text: Text Generation from Keywords for a Low-Resource Language”, advised by G. M. Shahariar Shibli. The work focused on keyword-driven Bangla text generation and was later published at LREC 2026.
- Contributed to final-year research as a research assistant on Bangla NLP, leading to the BIM 2023 publication “Rank Your Summaries: Enhancing Bengali Text Summarization via a Ranking-Based Approach”. This work shaped my research direction toward low-resource generative AI, reproducible evaluation, and practical NLP systems.
Core coursework
2018 — 2023Algorithms & Data Structures, Operating Systems, Distributed Database Systems, Computer Networks, Machine Learning, Pattern Recognition, Digital Image Processing, Artificial Intelligence.
Media
Ahsanullah University of Science and Technology
Convocation day marked the formal close of my undergraduate chapter at AUST and a quiet reminder of the research and engineering work that started there.
How I Think About Engineering
I gravitate toward problems at the boundary of correctness and performance: idempotent job queues, atomic Redis operations, schema migrations under live traffic, consistent hashing. These are the decisions that determine whether a system survives scale. I want to understand the invariants — what must always be true — before I write the first line.
What I'm Looking For
For PhD: advisors working at the intersection of multilingual NLP, multimodal learning, and low-resource AI — particularly those who value rigorous evaluation and reproducible research. For industry: backend engineering at genuine scale — real distributed systems, serious oncall culture, and space to do things right rather than just fast.