Magna Hire
Website:
magnahire.in
Job details:
🚀 Hiring: Founding Data Scientist – LLM Post-Training & Reinforcement Learning
Funding: $8M Seed-Funded AI Startup, backed by leading VCs
Experience: 7+ Years
Location: Gurgaon | 6 Days WFO
We're hiring for a well-funded AI startup building India's next-generation personal AI assistant — developing proprietary LLMs that learn from real-world interactions, user preferences, and task outcomes.
What You'll Own
- Fine-tune open-weight LLMs (Llama, Qwen, Gemma) using SFT, DPO, and RL.
- Build reward models and reinforcement learning pipelines to optimize agent decision-making.
- Develop training datasets from agent trajectories, tool interactions, and user preferences.
- Design LLM evaluation frameworks for reasoning, tool use, and task completion.
- Own distributed training, inference optimization, and production deployment.
Must-Have Skills
LLM Post-Training, SFT, DPO, RLHF/GRPO, PyTorch, LoRA/QLoRA, FSDP/DeepSpeed, Multi-GPU, vLLM/SGLang.
Why Join?
Work directly with experienced founders from Zomato, IDFC FIRST Bank, Kapture CX & Microsoft, building proprietary AI models from zero to scale.
📩 Referrals welcome!
Hiring in partnership with Magna Hire.
Click on Apply to know more.