Available for opportunities

Tawhid Ahmed

Software Engineer · AI/ML & Full-Stack

I build production web applications and intelligent systems — from Vue.js and Spring Boot enterprise platforms to RAG pipelines, agentic AI workflows, and computer vision.

AI / LLM in ProductionSpring BootVue 3Capital Markets
Tawhid Ahmed

About

Professional background and technical focus

I am a software engineer specializing in AI/ML and full-stack development. At Proficient Information Systems, I deliver features across three enterprise platforms — a DSE capital markets OMS, a Stellar-based remittance wallet, and SuperCards task-management SaaS — from database schema and Spring Boot services through Vue 3 interfaces, and own embedded AI / LLM capabilities end to end.

On the OMS, I built trading-terminal UI with live WebSocket feeds and owned AI features including a conversational trading assistant (MCP tool-use, Azure OpenAI) and per-symbol technical analysis. On SuperCards, I designed and built Task AI Coach — a FastAPI + LangGraph service with tenant-scoped RAG, view-scoped tools, and a production chat widget.

Alongside production work, I build open-source RAG and agentic AI systems, computer vision pipelines, and Spring Boot microservices — including Bangla OCR at ~96% validation accuracy and TaskPilot-AI with full-stack AI integration.

Skills

Technologies and tools I use day to day

Machine Learning & AI

TensorFlowPyTorchKerasScikit-learnModel TrainingData Preprocessing

NLP & LLMs

Hugging FaceLangChainLangGraphRAGFAISSMCPAzure OpenAISentenceTransformersPrompt Engineering

Computer Vision

OpenCVCNNsImage PreprocessingSegmentationTransfer LearningData Augmentation

Programming & Backend

Vue 3QuasarPiniaTypeScriptJavaScriptHTMLCSSPythonJava 21FastAPISpring Boot 3Spring CloudSQLPostgreSQLREST APIs

Data & Tools

NumPyPandasMatplotlibJupyter Notebooks

DevOps & Tools

DockerGitRedisRabbitMQCI/CDMLflowWebSocketsOAuth2JWT

Experience

Professional roles and impact

Junior Software Engineer

Sep 2024 – Present

Proficient Information Systems · Dhaka, Bangladesh

Deliver full-stack features across three ongoing enterprise platforms — a capital markets OMS, a cross-border remittance wallet, and an internal task-management SaaS — working from database schema through Spring Boot services to the Vue 3 interface, and own the embedded AI / LLM features end to end, with tests, API documentation, and code review throughout.

Programmer Intern

May 2024 – Aug 2024

Proficient Information Systems · Dhaka

Built responsive web applications using Vue.js, Quasar, HTML, CSS, and JavaScript, improving UI performance and usability. Assisted in small-scale project development, implementing new features and applying best practices for performance optimization. Collaborated with senior developers, gaining exposure to real-world software design patterns and agile practices.

Industrial Trainee

Nov 2020

BRACNet Limited · Dhaka

Completed 15-day intensive industrial training, gaining hands-on exposure to Data Center Services, IoT, Cybersecurity, ISP Technology, and Cloud Applications. Shadowed industry professionals and applied acquired technical skills to practical tasks.

Education

Academic background

M.Sc. in Computer Science & Engineering

Only Thesis Left

Bangladesh University of Engineering and Technology (BUET) · Dhaka

CGPA: 3.58. Advanced studies in computer science and engineering with focus on machine learning and software engineering.

B.Sc. (Eng.) in Information and Communication Engineering

2022

Noakhali Science and Technology University (NSTU) · Noakhali

CGPA: 3.66. Comprehensive education in information and communication engineering with hands-on experience in software development and machine learning projects.

Professional Work

Three enterprise platforms at Proficient Information Systems — full-stack delivery and embedded AI / LLM features

Capital Markets OMS & Trading Platform

Sep 2024 – Present

Proficient Information Systems · Investar BD

Junior Software Engineer

Spring Boot microservices platform for Dhaka Stock Exchange brokerage operations, covering the full trading lifecycle: order entry, pre-trade risk controls, FIX exchange execution, and real-time market data. Live at oms.investarbd.com.

  • Developed order-lifecycle, client-management, and pre-trade restriction endpoints across the OMS and CAT services within a nine-service microservices architecture, and added message consumers handling execution reports and order-status events.
  • Built the trading terminal front end — order entry, blotter, watchlist, and portfolio — streaming live order, execution, and market-data updates into the UI over WebSockets.
  • Owned the platform's AI features: a conversational trading assistant with LLM tool-use over MCP and streamed responses, and per-symbol technical analysis computing 30-day RSI, SMA, and VWAP indicators.

Java 21 · Spring Boot 3 · Spring Cloud · RabbitMQ · PostgreSQL · Redis · QuickFIX/J · STOMP WebSockets · OAuth2 · JWT · Azure OpenAI · MCP · Vue 3 · Quasar · Pinia · Docker

Cross-Border Remittance & Digital Wallet Platform

Sep 2024 – Present

Proficient Information Systems

Junior Software Engineer

Fiat ↔ USDC remittance platform integrating the Stellar blockchain and the MoneyGram anchor, backed by a double-entry ledger and a KYC compliance workflow.

  • Implemented REST endpoints and service-layer logic across the user, transaction, KYC, and ledger modules, contributing to a double-entry ledger with idempotent postings and a transaction state machine spanning 20+ deposit, withdrawal, transfer, hold, and reversal states.
  • Built the MoneyGram deposit and withdrawal integration over Stellar SEP-10 / SEP-24, with scheduled jobs polling provider status and reconciling inbound USDC payments streamed from Horizon.
  • Developed the wallet front end — authentication, dashboard, transactions, and KYC — including secure identity-document upload through backend presigned URLs.

Java 21 · Spring Boot 3.5 · Spring Modulith · PostgreSQL · Stellar SDK · Horizon · OAuth2 · JWT · Vue 3 · TypeScript · Quasar · Pinia · Docker

Internal production codebase

SuperCards — Task AI Coach

Sep 2024 – Present

Proficient Information Systems · Internal task-management SaaS

Junior Software Engineer · designed and built AI end to end

Conversational assistant for a multi-tenant task-management SaaS (SuperCards) that grounds an LLM in live backend data — giving management insight into task states and workload distribution, and helping individual users track and improve their own performance. Live at app.supercards.ca.

  • Designed and built the assistant as a FastAPI + LangChain / LangGraph microservice that grounds LLM answers in live REST-API data through a bounded tool-calling loop, removing an entire class of hallucinated figures.
  • Implemented nine domain tools behind a view-scoped registry so each page exposes only the relevant tools, with deterministic pre-LLM logic and scope labelling to keep reported totals accurate.
  • Built tenant-scoped RAG for tag resolution using per-tenant vector indexes with hybrid literal and semantic retrieval, and shipped the chat widget with persisted per-view threads and server-driven charts.

Python · FastAPI · LangChain · LangGraph · FAISS · Sentence-Transformers · Groq · Ollama · Laravel REST API · Vue 3 · Quasar · Pinia · ApexCharts

Open Source & Research

Independent AI/ML projects and published research implementations

Research RAG system

Research_Rag

Research-oriented RAG system for querying and synthesizing information from academic papers and research documents. Implements document ingestion, embedding pipelines, and retrieval-augmented generation for accurate, context-grounded answers.

PythonRAGLangChainFAISSFastAPIEmbeddings
GitHub
Agentic AI with LangGraph

Agentic AI with LangGraph

Agentic AI project built with LangGraph for orchestrating multi-step LLM workflows. Implements stateful agent graphs, tool calling, and structured reasoning pipelines for complex task automation.

PythonLangChainLangGraphLLMsAgentic AIFastAPI
GitHub
TaskPilot-AI Todo Assistant

TaskPilot-AI

AI-powered Todo app with secure per-user Spring Boot microservices and a Python AI agent using RAG (FAISS + SentenceTransformers) for context-aware reasoning. LangChain + LangGraph orchestrate multi-turn conversations, agent actions, and backend integration.

JavaSpring BootMicroservicesJWTPostgreSQLPythonFastAPILangChain+3
GitHub
Bangla OCR handwritten recognition

Bangla OCR

End-to-end system for offline Bangla handwritten character recognition, including compound characters. Implements image preprocessing, text segmentation, and CNN-based character classification with rigorous evaluation using precision, recall, F1-score, and confusion matrices.

  • ~96% validation accuracy on handwritten Bangla character recognition
PythonTensorFlowOpenCVNumPyPandasCNNComputer Vision
GitHub

Publications

Research and academic work

A Unified BERT-CNN-BiLSTM Framework for Simultaneous Headline Classification and Sentiment Analysis of Bangla News

Raquib, M., Akash, M.M., Ahmed, T., Murad, S.A., Prity, F.S., Hossain, M.A., Polok, A.P., and Rahimi, N.

arXiv preprint

2025

View on DOI

Offline Handwritten Character Recognition Including Compound Character from Scanned Document

Ahmed, T., Uddin, M., Khan, M.A.R. and Hasan, A.R.M.

Asian Journal of Research in Computer Science

Volume 14, Issue 42022Pages 119–129

View on DOI

Latest Writing

Articles from my Hashnode blog

View all →

No blog posts found yet. Check back soon or visit my Hashnode blog.

Visit Hashnode →