We engineer AI systems, web platforms, ML pipelines, and business software for production.
Somokolon Labs takes products from first sketch to a system real users rely on: designed, built, deployed, and monitored by one senior team.
Who we are
Somokolon Labs is a senior-led studio in Dhaka. We take ideas from a whiteboard to software people rely on: AI that is measured, not guessed, web apps that stay fast, and infrastructure you can actually see into.
Shahriar Ahmed Seam
Founder, Somokolon Labs
- Products shipped, every one with a live demo
- 20
- Technologies across our production stack
- 39
- Practice areas, from applied AI to infrastructure
- 4
What we do
Four practices. One team that ships.
From the model to the interface to the infrastructure underneath, the same people carry the work from the first call to production.
All servicesAI & LLM Engineering
RAG & LLM Applications
Retrieval-augmented systems with hybrid search, reranking, and evaluation — grounded answers over your own data.
Agentic Workflows
Multi-step agents that plan, use tools, and act safely, with human-in-the-loop where it matters.
Model Fine-Tuning
Task-specialized open models (LoRA/QLoRA) with reproducible evaluation against real baselines.
On-Device & Edge AI
Quantization and ONNX conversion for privacy-preserving, offline-first inference on constrained devices.
Web & Full-Stack Development
Custom Web Applications
Tailored products built from the ground up around your workflow, not a template.
Redesign & Modernization
Rework dated interfaces and legacy stacks into fast, maintainable, responsive apps.
APIs & Integrations
REST and GraphQL APIs, third-party integrations, and clean service boundaries.
Business Systems
Practical internal tools and POS/ERP-style systems for real operational needs.
Cloud & MLOps
Deployment & CI/CD
Automated build-test-deploy pipelines with GitHub Actions and container registries.
Model Serving & MLOps
Scalable inference serving, drift monitoring, and retrain pipelines for ML in production.
Observability
Metrics, logging, and tracing so you can see how systems behave under real load.
Event-Driven Systems
Decoupled, fault-tolerant services with queues and message streams that scale independently.
Quality Assurance & System Design
Automated Testing
Unit, integration, and end-to-end test suites wired into your CI pipeline.
AI/LLM Evaluation
Systematic evaluation harnesses for RAG and LLM systems — faithfulness, relevance, and cost.
Architecture Review
Design reviews focused on scalability, fault tolerance, and maintainability.
Performance & Load
Profiling and load testing to find bottlenecks before your users do.
Selected work
Products we designed, built, and run.
Every screenshot is the real interface, captured from the running product. Open any of them and try it yourself.
Industries
Industries we've built for
Every sector here has a working product behind it that you can open and try.
Retail
Point of sale that keeps trading when the network drops, and stock intelligence across locations.
Built
Banking & fintech
Double-entry ledgers with row-level locking, so concurrent transfers can't corrupt a balance.
Built
Healthcare & fitness
Telemedicine and encrypted records built for clinics here, plus on-device movement analysis.
Built
Logistics
Live vehicle tracking, geofencing in the database, and optimised routes for the day's stops.
Built
Agriculture
Crop disease detection that runs in the browser on a phone, with no signal needed in the field.
Built
Enterprise operations
HR, CRM, inventory, finance, and projects in one system, with AI that runs on your own hardware.
Built
Smart cities
Crowd and traffic counting from ordinary camera feeds, on CPU hardware rather than GPUs.
Built
AI & engineering teams
Model gateways, agent pipelines, and codebase search for teams putting LLMs into their products.
Built
Is your industry here?
If it isn't yet, tell us about the problem. That's how most of these started.
Let's talk
How we work
From first call to running system.
Four stages, one team, and working software at every checkpoint, so you always know where the project stands.
01Step 1 of 4
Discovery
We start with your problem and constraints — what success looks like, and what it will take to get there.
02Step 2 of 4
Plan & Estimate
A clear scope, milestones, and a realistic timeline, so you know what you're getting and when.
03Step 3 of 4
Build & Deploy
Iterative delivery with working software at each step — containerized and shipped, not left in a notebook.
04Step 4 of 4
Support & Iterate
Once it's live, we monitor, maintain, and improve it under real-world use.
Book a discovery callTech ecosystem
The stack behind what we ship.
39 tools in production across our products today. We pick them for the system in front of us, not for the trend.
- PyTorch
- Hugging Face Transformers
- PEFT / LoRA
- LangGraph
- Ollama
- ONNX Runtime
- scikit-learn
- OpenCV
Model training, fine-tuning, and inference — from transformers to diffusion and on-device CNNs.
Let's talk
Have a problem worth solving?
Tell us what you're building. We'll help you scope it, build it, and ship it.