Career Profile
AI engineer with several years of software engineering experience across generative AI, healthcare platforms, microservices, and cloud systems. Combines hands-on AI development with a strong foundation in Python, .NET, and distributed software architecture. Currently pursuing an MSc in AI & Language at Stockholm University and seeking Research Engineer, AI Engineer, or AI Software Engineer opportunities.
Experiences
- Developed Hear-Me-Out, a research platform for controlled studies comparing conversational AI responses to natural and converted voices.
- Built React/TypeScript participant and researcher interfaces with FastAPI services for study configuration, counterbalanced assignments, and questionnaires.
- Engineered an Opus-over-WebSocket audio pipeline integrating NVIDIA PersonaPlex with MeanVC/X-VC for real-time voice conversion, plus Seed-VC for offline conversion.
- Implemented server-controlled voice switching and captured original and converted audio, transcripts, and session events for WER, speaker similarity, and UTMOS analysis.
- Migrated speech inference to local GPUs and deployed the services with Docker, isolated Python environments, and HTTPS/WSS.
- Led generative AI projects using Llama.cpp, LlamaIndex, Whisper, OpenAI, and Mistral.
- Built an AI-assisted mainframe code modernization platform for code generation, testing, documentation, dependency graphs, and migration plans.
- Developed a multi-tenant Blazor clinic portal and continued .NET WPF kiosk applications for appointments, check-in, and payments; added FlaUI regression tests.
- Built Python microservices, Dockerized code-execution sandboxes, and distributed processing with Celery and Redis.
- Designed a graph architecture for code dependencies and stored transformation artifacts in MinIO; built a FastAPI/SSE monitoring dashboard with Grafana and Langfuse tracing.
- Designed event-driven microservices using Azure Functions and Service Bus, with Redis Cache and Azure Blob Storage.
- Integrated Cambio Cosmic, DIPS, Semble, and Dentally EHR systems with appointment and payment features; developed related WPF desktop applications.
- Built .NET Core and PostgreSQL healthcare services, including payment, invoicing, and debt-collection workflows.
- Managed API access through Azure API Management and CI/CD pipelines through Azure DevOps.
- Built data systems with Cosmos DB, MSSQL, and Entity Framework, and maintained technical documentation.
- Worked closely with Project Managers and other developers to define parameters and plan detailed specifications with clear project deliverables and timelines to meet strict deadlines.
- Developed back-end Web services in ASP.NET core with microservices architectures.
- Developed reactive web application with Rx.NET and Blazor
- Worked with the latest cutting-edge technologies like Docker,Kubernetes,GitHub-Actions etc
- Designed client-side code with Angular, Rxjs, specifically known as reactive web app
- Collaborated in integrating diverse stack web UI applications featuring web components from React, Angular etc.
- Implemented real-time communication and data transmission through SignalR and server-sent events.
- Developed high-performance API with .Net Core
- Developed Windows and Linux worker services
- Contributed to a map based mobile application in the Flutter
- Introduced and implemented a successful DevOps pipeline for the first time, resulting in a significant reduction in deployment failures, bugs, and overall deployment time.
Certifications
Projects
GpBot
| Python, PyTorch, Streamlit, Docker, BM25, BiLSTM
Source code A bilingual FAQ chatbot for Grameenphone that answers customer questions in English and Romanized Bangla.
Source code A bilingual FAQ chatbot for Grameenphone that answers customer questions in English and Romanized Bangla.
- Trained a BiLSTM classifier to route queries into 14 intents and used BM25 to retrieve relevant FAQ answers.
- Built a Streamlit chat interface with conversation history.
Terminal
| .NET Core, WPF, Blazor, Azure, FlaUI
A multi-tenant healthcare platform for clinic appointments, patient management, billing, and self-service.
A multi-tenant healthcare platform for clinic appointments, patient management, billing, and self-service.
- Built a shared backend, WPF patient kiosk, and Blazor portal for clinic staff across multiple organizations.
- Extended workflows from the Nordic region to the UK and integrated Cambio Cosmic, DIPS, Semble, and Dentally.
- Added automated desktop UI testing with FlaUI.
MeetSum
| Python, NextJS, Runpod, Docker, CosmosDB, Llamacpp, Whisperx, Mistral-7b
MeetSum is an AI-powered assistant that specializes in recording meeting audio, diarizing, and transcribing it using advanced open weight finetuned-LLM served in private data center. It alleviates the need for manual note-taking, allowing participants to stay more focused during discussions. MeetSum enhances meeting efficiency by automatically generating concise, accurate summaries and actionable minutes. This tool seamlessly integrates with any meeting platforms, ensuring a smooth user experience.
MeetSum is an AI-powered assistant that specializes in recording meeting audio, diarizing, and transcribing it using advanced open weight finetuned-LLM served in private data center. It alleviates the need for manual note-taking, allowing participants to stay more focused during discussions. MeetSum enhances meeting efficiency by automatically generating concise, accurate summaries and actionable minutes. This tool seamlessly integrates with any meeting platforms, ensuring a smooth user experience.
- Led GenAI project development, overseeing task management, ensuring timely delivery, and coordinating efforts across the development team.
- Developed advanced transcription and summarization features using fine-tuned large language models such as OpenAI’s Whisper and Mistral-7B.
- Implemented full-stack development, developing audio recording features with RecordRTC and utilizing Next.js for front-end development to create a seamless user interface that works with any meeting platform.
- Integrated Azure Blob Storage for efficient handling and storage of large audio datasets.
- Used Llamacpp to serve and query LLM efficiently, successfully serving quantized models to increase GPU memory efficiency while maintaining relevant results.
- Optimized scalability and performance through serverless architecture using Runpod and Docker
Jengax
| .NET Core, Azure Function, CosmosDB, Redis, Azure Service Bus
JengaX is here to solve crypto's biggest problem: making it easy to use for everyone JengaX API is an API product that facilitates crypto buying and selling using local fiat currency. The product was initially launched for the African market and it eventually will operate in europe. The API is a cloud native project that can scale to any demand and operates as a serverless system. This model allows for maximum performance and cost efficiency at the same time.
JengaX is here to solve crypto's biggest problem: making it easy to use for everyone JengaX API is an API product that facilitates crypto buying and selling using local fiat currency. The product was initially launched for the African market and it eventually will operate in europe. The API is a cloud native project that can scale to any demand and operates as a serverless system. This model allows for maximum performance and cost efficiency at the same time.
- Contributed to develop subscription-based microservice API with Azure APIM ,Azure Function, Azure Service Bus and Azure Active Directory
- Contributed to apply rate limit in the current system synchronized with third party systems
- Developed token bucket using semaphore to avoid duplicate auth request to external system
- Developed CI/CD pipeline with Azure Devops and Powershell to automate the deployment in dev,uat,test and prod environment
- Contributed to develop Iaac for the project with help of Bicep
- Developed and designed efficient database containers, aligned with intricate business logic, to optimize the SAAS application’s performance and scalability through sharding.
- Developed notification services to notify important messages through webhooks and socket URL with SignalR
Geologiq
| .Net Core, Angular, RxJS, Docker, Azure Devops
GeologiQ is a DIGITAL TWIN of an entire oilfield loaded with data and tools that will assist DRILLING & WELL teams in every part of the WELL DELIVERY PROCESS.The key to a successful well lies in the details. The combination of 3D and 2D makes it possible to TRANSFER EXPERIENCE and extract REFERENCE WELL data faster in a game like 3d environment.GeologiQ models are based on REAL DATA - either manually gathered or by integration with the client 3rd party API's.
- Contributed to development of microservice-based api using .NET Core
- Designed and developed reactive front-end application with Angular,RxJS and Blazor
- Implemented continuous delivery using Azure Devops to deploy the system upon new release
- Collaborated closely with clients to efficiently manage and resolve new tasks in Azure DevOps Kanban board, ensuring seamless workflow and timely resolution of issues.
Wellbook
| Angular, RxJS, Typescript
Wellbook is a project planning software built for specifically well industry,That helps well engineers to assign tasks,track progress and assist to take decisions by visualizing different graphs based on data produced by Geologiq.The initial development was based on a desire to digitize a process normally managed in spreadsheets. Wellbook is made by drilling engineers for drilling engineers. It can easily be adapted to client work processes like well planning, interventions, rig intakes etc. The key in Wellbook are the activities (eg. tasks, meetings, decision gates). Each activity is linked to input, output (results), documentation, RACI roles, decisions, actions, minutes of meeting, comments, progress and of course a detailed description of how to perform the task.The architecture of wellbook makes it possible to optimize each part of the planning process in a digitial way, step by step. Orchestrating the movement of data instead of documents.
- Built the backend .NET Core API for the project.
- Integrated the API with an existing Angular application using micro-frontend architecture, leveraging web components.
- Designed and developed a reactive front-end application with Angular and RxJS.
- Created various project management components.
Publications
Daffodil International University, 2019-12-10
Technical Skills
Languages
- C#
- TypeScript
- JavaScript
- PowerShell
- Python
- R
- Bicep
Frameworks
- ASP.NET Core
- .NET Core
- FastAPI
- Angular
- Blazor
- RxJS
- Entity Framework
- Microsoft Graph
- SignalR
- XUnit
DevOps & IaC
- Git
- Docker
- TravisCI
- GitHub Actions
- Azure DevOps
- Bicep
- Pulumi
Design Patterns
- SOLID
- Singleton
- Factory
- MediatR
- CQRS
Databases
- MSSQL
- PostgreSQL
- Cosmos DB
- Redis
- Pinecone
- Neo4j
AI
- PyTorch
- scikit-learn
- NumPy
- pandas
- Matplotlib
- dplyr
- ggplot2
- tidyverse
- OpenAI
- vLLM
- LangChain
- LlamaIndex
- AutoGen
- CrewAI
- Llama.cpp
- PEFT
- RAG
- MCP
Observability
- Azure Application Insights
- Grafana
- Langfuse
- Weights & Biases
Authentication
- Auth0
- Microsoft Entra ID
- OAuth 2.0
- OpenID Connect
- JWT
- RBAC
Cloud
- AWS
- Azure
- Kubernetes