Building AI that speaks Bangla, runs on your servers and answers to you
Hackules is an AI research lab building Bangla-native, low-resource and sovereign AI — from models like Sangbad-SLM to the on-shore infrastructure that runs them — together with governments, researchers and enterprises in Bangladesh and beyond.
Partners & collaborators
2021
Founded in Dhaka
0
Researchers & Engineers
0+
Research & Pilot Projects
0+
Partner Organisations
What the Lab Works On
Four connected research areas, one team: we build Bangla models, prove them in the field, make them run on sovereign infrastructure and grow the partnerships that put them to work.
Bangla AI
Language, speech and multimodal models built from the ground up for Bangla and other low-resource settings.
We train our own models rather than only fine-tuning someone else's. Our flagship Sangbad-SLM, a 0.8B-parameter Bangla news language model, is built on a corpus we collected and cleaned ourselves, and it runs on modest hardware.
- Sangbad-SLM & Bangla Language Models
- Bangla Corpus Engine
- Low-Resource Training Methods
- Speech & Dialect Research
- Sovereign Benchmarks
- Efficient Inference
Applied AI Research
Turning research into working systems for documents, vision, data and decision-making.
The lab's applied group takes models from the bench into the field: document intelligence for Bangla and English paperwork, computer vision, forecasting and retrieval-augmented assistants, each measured against real-world data.
- Document & Data Intelligence
- Computer Vision
- Predictive Modelling
- Retrieval & Generative Systems
- AI Evaluation & Safety
- Experimental Prototypes
Sovereign AI Infrastructure
On-shore, private infrastructure so institutions can run their own models on their own data.
Owning a model is only useful if you can run it. We research and build the inference pipelines, deployment blueprints and governance needed to operate AI on your own servers or private cloud, inside your borders.
- National Inference Pipeline
- On-Premise & Private Cloud
- MLOps & Model Monitoring
- Data Governance & Privacy
- Compute Strategy
- Open Standards & Interoperability
Innovation Programs & Partnerships
Joint research, pilots and capacity building with government, academia, NGOs and industry.
We work best alongside partners who bring a hard problem and real data. Our programs range from short pilots and proofs of concept to multi-year joint research, with training so partner teams can carry the work forward themselves.
- Joint Research Projects
- Pilots & Proofs of Concept
- AI Readiness & Strategy
- Capacity Building
- Data Partnerships
- Long-Term Support
Research first,
rooted in Bangladesh
We are not a reseller of someone else's API. We collect the data, train the models, write the benchmarks and design the infrastructure, so every system we put in a partner's hands is one we understand end to end and one they can own.
Our own models
Sangbad-SLM and the models that follow it are trained from scratch on corpora we built, not fine-tuned on borrowed weights.
Bangla-native, low-resource
We specialise in the languages, dialects and compute budgets that the largest labs overlook.
Sovereign by design
Models sized to run on-premise or in a private cloud, with data residency, access policies and audit trails built in.
Proven with institutions
Research that has already run inside government agencies, NGOs and enterprises in Bangladesh and beyond.
Research that starts with Bangla
The lab is organised around one flagship model and four research pipelines focused on Bangla-native, low-resource and sovereign AI — so every system we put into the world is one we understand end to end.
Sangbad-SLM
0.8B parameters · Bangla news language model
Bangladesh's first purpose-built Bangla news language model, trained from the ground up on our own Bangla news corpus. Designed to run on modest hardware with low inference costs, it outperforms larger general-purpose models on Bangla news comprehension.
- ✓ Trained on a proprietary Bangla news corpus
- ✓ Runs on-premise on modest GPU/CPU hardware
- ✓ Low inference cost for national-scale deployment
Bangla Corpus Engine
Large-scale collection and cleaning of Bangla text, speech and dialect data — the foundation for every model we train.
Low-Resource LLM Training
Fine-tuning and pretraining techniques optimised for limited compute, so strong models stay affordable to build and run.
Sovereign Benchmarks
Evaluation suites built for local contexts — government, healthcare and education — so you can trust the numbers.
National Inference Pipeline
An on-shore infrastructure blueprint for public institutions: your models, your data, your servers.
Founded in Dhaka as an AI research lab
AWS Activate grant recipient (2023)
Backed by BYLC Ventures and a Bangladesh accelerator
Where Our Research Lands
We work in high-stakes, often under-served domains where language, data sovereignty and limited compute decide whether AI helps at all.
Government & Public Sector
National-scale Bangla language services, document digitisation and sovereign inference for citizen-facing institutions.
Media & Journalism
Sangbad-SLM grew out of Bangla news: comprehension, summarisation and newsroom tools that understand local reporting.
Education & Research
Bangla reading and learning tools, curriculum-scale content analysis and joint research with universities.
Healthcare
Clinical document intelligence and privacy-first models that stay inside the hospital's own infrastructure.
Finance & Inclusion
Bangla document understanding, risk and anomaly models for banks, fintechs and mobile-money providers.
Development & Field Programs
Field-data intelligence, speech interfaces and analytics for NGOs and large programmes reaching people offline.
From research question to running system
A transparent, evidence-first way of working that keeps partners in control and ends with them owning the result.
Frame
Define the problem with your team, audit the data, and agree the evaluation that will decide success.
Research
Experiments, model training and benchmarking against baselines, with results shared openly as we go.
Pilot
A time-boxed prototype in the field with real users and real data, and an honest decision on what to scale.
Transfer
Deployment on your infrastructure, monitoring, and training so your team can operate and extend the system.
Tools We Build With
Open, well-supported tooling chosen so that partners can reproduce our work and run it on hardware they already have.
Models & Training
Data & Evaluation
Compute & Deployment
Organisations We Have Worked With
Have a problem AI could solve?
Tell us about your research question, your data or the people you serve, and the lab will get back to you within one business day.
Talk to the lab