G
Gnani.ai operates a voice-native AI platform designed to automate customer-facing workflows at scale—processing 30M+ voice interactions daily across voice, chat, and omnichannel surfaces. The tech stack reveals a GPU-heavy, LLM-first engineering culture: PyTorch + DeepSpeed + Megatron-LM for distributed training, NVIDIA A100/H100 infrastructure, and active investment in multilingual SLM pre-training (7B–70B+ models). Hiring is engineering-led and accelerating, with product leadership (Director-level) and sales roles being added in parallel—suggesting a transition from product-market fit in BFSI/contact-center use cases toward scaling globally.
Tech Stack 28 technologies
What gnani.ai Is Building
◆Challenges
- Scaling llm training to 70b+ models
- Optimizing gpu utilization
- Cleaning multilingual datasets
- Meeting sales quotas
- Increasing adoption of voice ai solutions
- Scaling to global markets
- Building next generation human-ai collaboration infrastructure
▲Active Projects
- Data creation and cleaning for multilingual datasets
- Ai-based product implementations
- Ai-integrated product deliveries
- Llm pre-training pipelines (7b–70b+)
- Distributed training with pytorch distributed, deepspeed, megatron-lm, nvidia nemo
- Llm pre-training pipeline optimization
- Distributed training implementation
- 12–18 month product vision and roadmap for inya.ai, speech analytics, and agent assist
- Evolution of agentic ai capabilities
- Build next generation human-ai collaboration infrastructure
Hiring Activity
9 roles · 5 in 30d
Department
Seniority
Notable leadership hires: Product Director
Company intelligence
